Ocean thermohaline structure and dynamic height multi-scale inversion method

Abstract

The invention belongs to the technical field of marine environment parameter inversion and marine dynamics observation, and particularly relates to a marine thermohaline structure and dynamic height multi-scale inversion method, which comprises the following steps of: acquiring back-and-forth sound wave propagation time and seabed pressure data measured by a PIES arranged at the seabed of a target sea area; preprocessing the data to generate oblique pressure contribution sound propagation time data; on the basis of the high-temporal-spatial-resolution ocean mode data, temperature-salinity profile data corresponding to the PIES position are obtained, and the sound propagation time of the model is calculated; performing multi-scale dynamic signal separation on the model data and the PIES data; constructing a multi-scale gravity empirical mode mapping function; inverting the thermohaline profile and the dynamic height of each scale by using the mapping function; and superposing the results of each scale to generate a multi-scale synthesized total temperature salt structure and total power height. According to the invention, based on multi-scale joint inversion of PIES data and high-resolution numerical values, the capturing and inversion capability of sub-mesoscale and high-frequency dynamic processes is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of ocean environmental parameter inversion and ocean dynamics observation, and specifically relates to a multi-scale inversion method for ocean temperature-salinity structure and dynamic height. Background Art

[0002] The temperature, salinity (referred to as the temperature-salinity structure) and dynamic height of the ocean are important physical characteristics of ocean dynamic processes. The temperature-salinity structure directly reflects the density characteristics of seawater, and the dynamic height is formed by the vertical integration of the seawater density anomaly caused by the temperature-salinity structure. The two are closely related and jointly determine the three-dimensional flow field structure of the ocean. Accurately obtaining the multi-scale temperature-salinity structure and dynamic height is crucial for analyzing and predicting ocean circulation, climate change, energy and material transport, and local ocean ecological environment.

[0003] Currently, the observation of ocean temperature-salinity structure and dynamic height mainly relies on methods such as satellite remote sensing, shipborne observation equipment (such as CTD), Argo floats, underwater gliders, and seafloor mooring observation systems. Satellite altimeters have the advantage of long-term continuous global observation and have become an important means for studying large-scale and mesoscale ocean dynamic processes. However, due to the limitations of the spatial resolution and observation period of satellite altimeters (the spatial scale is generally greater than 100 km, and the time scale is generally greater than 10 days), it is difficult to effectively resolve sub-mesoscale (1 to 50 km) and high-frequency processes such as tides. Although shipborne CTD and Argo floats can provide accurate temperature-salinity profile data, the observation cost is high, and the spatial coverage is limited, making it difficult to continuously observe small-scale processes for a long time.

[0004] Pressure Inverse Echo Sounder (PIES) has been widely used in the observation and research of ocean temperature-salinity structure and dynamic height in recent years. However, the currently widely used traditional Gravity Empirical Mode (GEM) method usually constructs a static empirical mode mapping relationship with historical discrete observed temperature-salinity data, lacking time continuity and fine-scale information. Therefore, it is unable to effectively capture and invert the rapid changes of sub-mesoscale and high-frequency processes, resulting in limited inversion accuracy and scale resolution ability.

[0005] On the other hand, high-resolution numerical models can finely simulate the changes in the temperature-salinity structure at the mesoscale, sub-mesoscale, and tidal scales of the ocean, providing high-quality and continuous temperature-salinity structure simulation data for the study of high-frequency and sub-mesoscale dynamic processes. The fusion of this model data and PIES observation data can significantly enhance the scale resolution ability of ocean dynamic processes. However, at present, there is still a lack of an effective technical method to combine the temperature-salinity information of high-resolution models with PIES high-frequency observation data to establish a multi-scale mapping relationship that can accurately invert the temperature-salinity structure and dynamic height simultaneously.

[0006] Therefore, there is an urgent need in this field to propose a technical method for accurate, multi-scale joint inversion of ocean temperature and salinity structure and dynamic altitude. Summary of the invention

[0007] The purpose of the present invention is to provide a multi-scale inversion method for ocean temperature-salinity structure and dynamic altitude based on a pressure sensing inverse echo sounder (PIES). By innovatively constructing a multi-scale gravity empirical mode mapping function (MS-GEM-EMMF), a multi-scale joint inversion method based on PIES data and high-resolution mode data is realized, thereby effectively improving the inversion accuracy and scale resolution of the temperature-salinity structure and dynamic altitude, so as to overcome the defects of the above-mentioned existing inversion methods.

[0008] The technical solution adopted by the present invention to achieve the above-mentioned purpose is: a multi-scale inversion method for ocean temperature-salinity structure and dynamic height, comprising the following steps:

[0009] S1: Collect the round-trip sound wave propagation time and seabed pressure data measured by the PIES deployed on the seabed of the target sea area at a set frequency;

[0010] S2: Preprocess the data collected in step S1 to generate baroclinic contribution acoustic propagation time data with a resolution of no less than 1 hour ;

[0011] S3: Based on high temporal and spatial resolution ocean model data, obtain the temperature and salinity profile data corresponding to the PIES position near the target sea area and calculate the model sound propagation time ;

[0012] S4: performing multi-scale kinetic signal separation on the model data and PIES data;

[0013] S5: Based on the data of multi-scale dynamic signal separation, a multi-scale gravity empirical mode mapping function is constructed, and the statistical relationship between sub-scale sound propagation time and temperature-haline structure is established;

[0014] S6: Invert temperature-salinity profiles and dynamic altitude at each scale using multi-scale gravity empirical mode mapping function;

[0015] S7: Superimpose the results of each scale to generate the multi-scale synthetic total temperature and salt structure and total dynamic height.

[0016] The observation frequency of PIES is no less than once per hour to be suitable for sub-mesoscale and high-frequency process observations.

[0017] In step S2, the data is preprocessed, including the following steps:

[0018] S21: Use the 3σ criterion to eliminate outliers in the sound wave propagation time data;

[0019] S22: Correct the long-term drift error of the instrument by the linear regression method;

[0020] S23: Calculate the barotropic component using the seabed pressure data, and perform dynamic compensation on the acoustic propagation time through the following formula:

[0021] where, is the original propagation time, is the change in seabed pressure, is the reference density, is the acceleration due to gravity, and c is the sound speed.

[0022] In step S3, the high spatio-temporal resolution ocean model data satisfies the following parameters: Temporal resolution: 1 hour; Horizontal spatial resolution: ≤1 / 48°; Vertical resolution: 1 m within 0 - 200 m depth, 10 - 50 m within 200 - 1000 m depth; including physical processes of submesoscale eddies, fronts, and tidal forcing.

[0023] The step S3 is specifically as follows:

[0024] S31: Unify the temporal resolution of the model data and the PIES observation data using the sinc interpolation method;

[0025] S32: Traverse the loop to find the data with the nearest PIES deployment longitude and latitude in the model data, and use cubic spline hierarchical interpolation to the PIES deployment station to obtain the corresponding T / S data;

[0026] S33: Obtain the seawater density and the sound speed c according to the T / S data, and then obtain the model acoustic propagation time as:

[0027] where c is the sound speed, is the seawater density, g is the acceleration due to gravity, is the bottom reference pressure of the PIES station.

[0028] In steps S4 and S6, for the multi-scale dynamic signal separation, time-domain band-pass filtering processing is adopted, and the specific scale frequency band division includes but is not limited to: Mesoscale: period > 10 days; Submesoscale: period 1.5 days - 10 days; Diurnal tide: period 18 hours - 36 hours; Semidiurnal tide: period 10 hours - 18 hours.

[0029] The step S5 specifically includes the following steps:

[0030] S51: Data layering processing: Stratify the temperature-salinity data T / S separated by multi-scale by water depth, with each layer having a thickness of N meters; perform stratification processing independently for each scale band to ensure independent modeling of temperature-salinity profiles at different scales;

[0031] S52: For each layer of data, use the weighted least squares method to fit the acoustic propagation time and the cubic polynomial relationship of the corresponding temperature-salinity data T / S, that is:

[0032] where, ~ are the fitting coefficients and ϵ is the residual.

[0033] Allocate weights according to the confidence level of the temperature-salinity data, with high-confidence data having a higher weight in the fitting to improve the robustness of the mapping relationship.

[0034] S53: Cross-validation optimization: Divide the data into a training set and a validation set to verify the fitting effects of different polynomial orders; calculate the residual sum of squares (RSS) of each order model, and select the polynomial order with the smallest RSS as the final model; if the cubic polynomial is already optimal, directly adopt it; if there is no significant improvement in higher orders, avoid overfitting and keep the cubic polynomial form.

[0035] S54: Construct a multi-scale mapping function: Independently repeat steps S51 - S53 for each scale component, and finally form a layered-multi-scale mapping function library; complete the construction of the multi-scale gravity empirical mode mapping function.

[0036] The specific steps of step S6 are as follows:

[0037] S61: Perform multi-scale dynamic signal separation on the PIES baroclinic contribution acoustic propagation time data to obtain the propagation times at different scales ;

[0038] S62: Use the propagation times at different scales obtained in step S61 as input data and substitute them into the constructed mapping function to inversely obtain the temperature-salinity profiles at the corresponding scales;

[0039] S63: Calculate the corresponding sub-scale ocean dynamic height SH for the temperature-salinity fields obtained at each scale in step S62.

[0040] Calculate the corresponding sub-scale ocean dynamic height SH, that is:

[0041] where, is the dynamic height at each scale, is the reference depth of the PIES station, represents the seawater density calculated by inputting the T / S profile obtained by inversion into the Gibbs-SeaWater formula library, where, is the long-term average density, is the obtained density anomaly.

[0042] The step S7 is specifically as follows:

[0043] Sum by scale to obtain the comprehensive multi-scale temperature ( ), salinity ( ), and dynamic height ( ), that is:

[0044]

[0045]

[0046] where scale represents the scale component.

[0047] The present invention has the following beneficial effects and advantages:

[0048] 1. By innovatively proposing a multi-scale gravity empirical mode mapping function, the present invention realizes multi-scale joint inversion based on PIES data and high-resolution numerical model data, effectively overcoming the static empirical mode mapping relationship of traditional methods that only rely on historical discrete observation data, significantly improving the ability to capture and invert sub-mesoscale and high-frequency dynamic processes, and being able to effectively resolve ocean processes with a spatial scale as small as 2 km and a time scale as short as several hours.

[0049] 2. The method proposed by the present invention utilizes the high-frequency information contained in PIES observation data and the sub-mesoscale dynamic information in high-resolution numerical model data to achieve an organic fusion of the two data in statistical characteristics. Since the high-resolution numerical model can accurately simulate the statistical characteristics of the multi-scale temperature and salinity structure of the ocean, and has good consistency with the echo propagation time records of PIES high-frequency continuous observations in statistical terms, it ensures the accuracy and consistency of the inversion results at different scales.

[0050] 3. The method of the present invention can precisely extract and analyze the contributions of ocean processes at different scales to the dynamic height change, such as mesoscale (greater than 10 days), submesoscale (1.5 - 10 days), and high-frequency tidal scale (3 - 36 hours), significantly improving the resolution ability of different-scale ocean dynamic processes. This method can extract and restore important submesoscale and high-frequency signals that are filtered or ignored by traditional methods from PIES high-frequency observation data, effectively avoiding the loss of high-frequency detail information and improving the sensitivity of the inversion result to ocean dynamic changes.

[0051] 4. Compared with satellite remote sensing and shipborne equipment, the present invention has lower observation costs and higher spatio-temporal resolutions, and is particularly suitable for long-term, continuous, and multi-scale observations of ocean temperature-salinity structure and dynamic height. Utilizing the high-frequency observation ability of the PIES device and cooperating with multi-scale numerical simulation data of high-resolution models, long-term continuous and precise observations of ocean internal dynamic processes can be efficiently and economically achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a flow block diagram of the inversion method of the present invention.

[0053] Figure 2 It is an example of the inversion of submesoscale temperature anomaly in shallow waters of 0.5 - 3 days in this embodiment.

[0054] Figure 3 It is an example of the inversion of submesoscale temperature anomaly in shallow waters of 3 - 10 days in this embodiment.

[0055] Figure 4 It is an example of the inversion of mesoscale temperature above 10 days in this embodiment.

[0056] Figure 5 It is an example of the comprehensive multi-scale temperature inversion result in this embodiment.

[0057] Figure 6 It is an example of the true value in this embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0058] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0059] In the actual application process, the multi-scale inversion method of ocean temperature-salinity structure and dynamic height based on the pressure-sensing inverse echo sounder (PIES) is as shown in the following embodiments.

[0060] Embodiment

[0061] As Figure 1 shown, it is a flowchart of the inversion method of the present invention. A multi-scale inversion method for ocean temperature-salinity structure and dynamic height of the present invention is implemented based on a pressure-sensing inverse echo sounder (PIES) and a high-resolution model. The specific method includes the following steps:

[0062] S1: Obtain the round-trip acoustic wave propagation time measured by PIES and the seabed pressure data deployed on the seabed of the target sea area. The acoustic wave propagation time data and the bottom pressure data are collected at a frequency of 24 times per hour;

[0063] S2: Remove outliers from the acoustic wave round-trip propagation time data and the seabed pressure data, correct the long-term drift, and use the seabed pressure data to perform a quality load (i.e., barotropic component) correction process on the measured propagation time to process it into baroclinic contribution acoustic propagation time( ) data;

[0064] S21: Remove outliers in the acoustic wave propagation time data using the 3σ criterion;

[0065] S22: Correct the long-term drift error of the instrument by the linear regression method;

[0066] S23: Calculate the barotropic component using the seabed pressure data, and perform dynamic compensation on the acoustic propagation time through the following formula:

[0067] where, is the original propagation time, is the change in seabed pressure, is the reference density, is the acceleration due to gravity, and c is the sound speed.

[0068] S3: Use the high-resolution ocean numerical model data to loop through and find the data in the model data that is closest to the longitude and latitude where PIES is deployed. Use cubic spline hierarchical interpolation to the PIES deployment station to obtain the high spatio-temporal resolution temperature-salinity (T / S) profile data near the target sea area. Use the sinc interpolation method to unify the time resolution of the model data and the PIES observation data; and calculate the propagation time by the following formula :

[0069] where, c is the sound speed, which is calculated by the sound speed empirical formula that meets the usage condition range of the observation sea area through the T / S at the station obtained in S31, is the seawater density calculated from T / S, g is the acceleration due to gravity, is the bottom reference pressure of the PIES station.

[0070] Among them, the high spatio-temporal resolution ocean model data meet the following parameters: Temporal resolution: 1 hour; horizontal spatial resolution: ≤1 / 48°; vertical resolution: 1 meter within the depth of 0 - 200 meters, 10 - 50 meters within the depth of 200 - 1000 meters; including physical processes of submesoscale eddies, fronts and tidal forcing.

[0071] S4: Use time-domain band-pass filtering to separate multi-scale dynamic signals from the model data described in S3. The specific scale band division includes but is not limited to: mesoscale (>10 days), submesoscale (1.5 days - 10 days), diurnal tide (18 hours - 36 hours), semi-diurnal tide (10 hours - 18 hours).

[0072] S5: Use the T / S after scale separation and data. For each scale, use the cubic polynomial fitting method to construct a multi-scale gravity empirical mode mapping function (MS-GEM-EMMF) layer by layer;

[0073] S51: Data layer processing:

[0074] Layer the temperature-salinity data T / S after multi-scale separation by water depth, with each layer thickness of N meters; perform layer processing independently for each scale band to ensure independent modeling of temperature-salinity profiles at different scales;

[0075] S52: For each layer of data, use the weighted least squares method to fit the acoustic propagation time and the cubic polynomial relationship of the corresponding temperature-salinity data T / S, that is:

[0076] where, ~ are fitting coefficients, and ϵ is the residual.

[0077] Allocate weights according to the confidence level of the temperature-salinity data. High-confidence data occupy higher weights in the fitting to improve the robustness of the mapping relationship.

[0078] S53: Cross-validation optimization: Divide the data into a training set and a validation set to verify the fitting effects of different polynomial orders; calculate the residual sum of squares (RSS) of each order model, and select the polynomial order with the smallest RSS as the final model; if the cubic polynomial is already optimal, directly adopt it; if higher orders do not have significant improvement, avoid overfitting and maintain the cubic polynomial form.

[0079] S54: Construct a scale-separated mapping function: Steps S51 - S53 are independently repeated for each scale component, and finally a hierarchical - multi - scale mapping function library is formed; the construction of the multi - scale gravity empirical mode mapping function is completed.

[0080] S6: Perform multi - scale dynamic signal separation on the PIES baroclinic contribution sound propagation time ( ) data. The separation scale frequency band is the same as that described in step S4, and the propagation times at different scales ( ) are obtained.

[0081] S7: Take the propagation times at different scales ( ) obtained in step S6 as input data, substitute them into the mapping function constructed in step S5, and inversely obtain the corresponding scale temperature - salinity profiles.

[0082] S8: Calculate the corresponding scale - separated ocean dynamic height (SH) for the temperature - salinity fields obtained at each scale in step S7:

[0083] In the formula, is the dynamic height at each scale, is the reference depth of the PIES station, represents the seawater density calculated by inputting the T / S profile obtained by inversion into the Gibbs - SeaWater formula library. Among them, is the long - term average density, is the obtained density anomaly.

[0084] S9: Superimpose all the scale - separated temperature - salinity profiles and SH obtained in steps S7 and S8, that is, sum them by scale to obtain the comprehensive multi - scale temperature ( ), salinity ( ), and dynamic height ( ):

[0085]

[0086]

[0087] In the formula, scale represents the scale component.

[0088] Example 2:

[0089] The effects of the present invention will be described below in conjunction with the accompanying drawings of the examples.

[0090] Taking the temperature inversion in the shallow sea (with a water depth of about 111 meters) as an example, the high-resolution ocean numerical model MITgcmllc4320 is used to obtain the temperature and salinity (T / S) profile data with high spatio-temporal resolution near the target sea area, and a mapping function between the T / S time series corresponding to the PIES position and the sound propagation time is established. The MITgcm llc4320 model data has a time resolution of once per hour and a spatial resolution of 1 / 48°, and the vertical resolution is 1-50 meters in the depth range of 0-1000 meters, and its resolution meets the requirements of inversion.

[0091] The frequency bands for shallow sea inversion are divided into 0.5 days - 3 days ( Figure 2 ), 3 days - 10 days ( Figure 3 ), and more than 10 days ( Figure 4 ) to distinguish the shallow sea submesoscale and mesoscale processes. The superimposed comprehensive inversion result (such as Figure 5 ) compared with the true value (such as Figure 6 ) can accurately capture the high-frequency changes that cannot be captured by traditional inversion methods.

[0092] It can be seen from this that the method of the present invention can effectively extract and restore the important submesoscale and high-frequency signals filtered or ignored by traditional methods from the PIES high-frequency observation data, effectively avoiding the loss of high-frequency detail information and improving the inversion accuracy.

[0093] In summary, the method proposed by the present invention can not only significantly improve the inversion accuracy and scale resolution ability of the ocean temperature and salinity structure and dynamic height, but also show obvious advantages in terms of cost, efficiency, long-term continuous observation ability, etc., and can provide important technical support for multi-field research such as analyzing and predicting ocean circulation changes, climate variability, energy transport, and local ocean ecological environment.

[0094] In this specification, the present invention has been described with reference to its specific embodiments. The above embodiments are the preferred embodiments of this patent and are not used to limit the scope of implementation of the present invention. It should be noted that the present invention is not limited to the above specific implementation manners. Improvements, changes, combinations, substitutions, etc. made by those skilled in the art without departing from the principle of the present invention are all within the scope protected by the claims of the present invention.

[0095] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made by those skilled in the art once they learn the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention. Obviously, those skilled in the art can make various changes and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A multi-scale inversion method for ocean temperature-salinity structure and dynamic height, characterized in that: The following steps are involved: S1: Collect the round-trip sound wave propagation time and seabed pressure data measured by the PIES deployed on the seabed of the target sea area at a set frequency; S2: Preprocess the data collected in step S1 to generate baroclinic contribution acoustic propagation time data with a resolution of no less than 1 hour ; S3: Based on high temporal and spatial resolution ocean model data, obtain the temperature and salinity profile data corresponding to the PIES position near the target sea area and calculate the model sound propagation time ; S4: performing multi-scale kinetic signal separation on the model data and PIES data; S5: Based on the data of multi-scale dynamic signal separation, a multi-scale gravity empirical mode mapping function is constructed, and the statistical relationship between sub-scale sound propagation time and temperature-haline structure is established; S6: Invert temperature-salinity profiles and dynamic altitude at each scale using multi-scale gravity empirical mode mapping function; S7: Superimpose the results of each scale to generate the multi-scale synthetic total temperature and salt structure and total dynamic height.

2. The multi-scale inversion method of ocean temperature-salinity structure and dynamic height according to claim 1 is characterized in that: The observation frequency of PIES is no less than once per hour to be suitable for sub-mesoscale and high-frequency process observations.

3. The method for multi-scale inversion of ocean temperature-salinity structure and dynamic height according to claim 1, characterized in that: In step S2, the data is preprocessed, including the following steps: S21: Use the 3σ criterion to eliminate outliers in the sound wave propagation time data; S22: Correct the long-term drift error of the instrument by linear regression method; S23: Use the seabed pressure data to calculate the positive pressure component and dynamically compensate the sound propagation time using the following formula: ； in, is the original propagation time, is the seafloor pressure change, is the reference density, is the acceleration due to gravity, and c is the speed of sound.

4. The method for multi-scale inversion of ocean temperature-salinity structure and dynamic height according to claim 1, characterized in that: In step S3, the high temporal and spatial resolution ocean model data satisfies the following parameters: Temporal resolution: 1 hour; horizontal spatial resolution: ≤1 / 48°; vertical resolution: 1 meter within a depth of 0-200 meters, 10-50 meters within a depth of 200-1000 meters; includes the physical processes of sub-mesoscale vortices, fronts and tidal forcing.

5. The multi-scale inversion method of ocean temperature-salinity structure and dynamic height according to claim 1 is characterized in that: The step S3 is specifically: S31: Use sinc interpolation to unify the temporal resolution of model data and PIES observation data; S32: Loop traversal to find the data of the nearest PIES deployment latitude and longitude in the pattern data, and use cubic spline hierarchical interpolation to the PIES deployment station to obtain the corresponding T / S data; S33: Obtain seawater density based on T / S data and the speed of sound c, and then the model sound propagation time is obtained as: ； Where c is the speed of sound, is the density of seawater, g is the acceleration due to gravity, is the bottom reference pressure at the PIES station.

6. The method for multi-scale inversion of ocean temperature-salinity structure and dynamic height according to claim 1, characterized in that: In step S4 and step S6, the multi-scale dynamic signal separation is performed by time domain bandpass filtering, and the specific scale frequency band division includes but is not limited to: Mesoscale: period > 10 days; submesoscale: period 1.5 days to 10 days; diurnal tide: period 18 hours to 36 hours; semidiurnal tide: period 10 hours to 18 hours.

7. The multi-scale inversion method of ocean temperature-salinity structure and dynamic height according to claim 1 is characterized in that: The step S5 specifically comprises the following steps: S51: Data layering processing: The multi-scale separated temperature and salinity data T / S are layered according to water depth, with each layer being N meters thick; each scale frequency band is layered independently to ensure that temperature and salinity profiles at different scales are independently modeled; S52: For each layer of data, the weighted least squares method is used to fit the sound propagation time The cubic polynomial relationship with the corresponding temperature-salinity data T / S is: ； in, ~ is the fitting coefficient, ϵ is the residual; Weights are assigned according to the confidence of temperature and salinity data. High-confidence data has a higher weight in the fitting process to improve the robustness of the mapping relationship. S53: Cross-validation optimization: Divide the data into training set and validation set to verify the fitting effect of different polynomial orders; calculate the residual sum of squares (RSS) of each order model, and select the polynomial order with the smallest RSS as the final model; if the cubic polynomial is already optimal, use it directly; if there is no significant improvement with a higher order, avoid overfitting and keep the cubic polynomial form; S54: Constructing the scale mapping function: Repeat steps S51 to S53 for each scale component independently, and finally form a layered-scale mapping function library; complete the construction of the multi-scale gravity empirical mode mapping function.

8. The method for multi-scale inversion of ocean temperature-salinity structure and dynamic height according to claim 7, characterized in that: The step S6 is specifically: S61: Acoustic propagation time for the baroclinic contribution to PIES The data is separated into multi-scale dynamic signals to obtain the propagation time of different scales. ; S62: The propagation times of different scales obtained in step S61 are As input data, substitute it into the constructed mapping function and invert the temperature-salinity profile of the corresponding scale; S63: Calculate the corresponding sub-scale ocean dynamic height SH for the temperature-salinity field obtained at each scale in step S62.

9. The method for multi-scale inversion of ocean temperature-salinity structure and dynamic height according to claim 8, characterized in that: In step S63, the corresponding sub-scale ocean dynamic height SH is calculated, that is: ； in, is the dynamic height of each scale, is the reference depth of the PIES station, represents the seawater density calculated by the Gibbs-SeaWater formula library using the T / S profile obtained by inversion, where: is the long-term average density, for The resulting density is abnormal.

10. The method for multi-scale inversion of ocean temperature-salinity structure and dynamic height according to claim 1, characterized in that: The step S7 is specifically: The comprehensive multi-scale temperature is obtained by summing the scales ,salinity With power height ,Right now: ； ； ； Among them, scale represents the scale component.

Images