Ocean data gridding method for correcting density reversal

Abstract

The invention discloses an ocean data gridding method for correcting density reversal, and relates to the technical field of ocean data assimilation. The method comprises the following steps: acquiring multi-scale ocean data; constructing a three-dimensional regular grid, calculating a three-dimensional related scale, and determining a gridding search radius; background error correlation parameters are calculated through the three-dimensional correlation scale; determining the square ratio of the observation error to the background error; calculating a three-dimensional optimal weight through background error related parameters and the square ratio of the observation error to the background error; carrying out arithmetic averaging on the multi-scale ocean data in the specified angle range of the analysis grid points to obtain a gridding background field; observation values and background estimation values of the observation grid points are obtained; correcting the analysis grid points through the observation values of the observation grid points, the background estimation values, the three-dimensional optimal weights and the gridding background field to obtain an analysis field; and carrying out iterative calculation on the multi-scale ocean data through the analysis field so as to carry out density reversal correction on the multi-scale ocean data.

Description

Technical Field

[0001] The present application relates to the technical field of ocean data assimilation, and particularly relates to a method for gridifying ocean data for correcting density inversion. Background Art

[0002] At present, gridification of data can obtain a large number of observation data that are evenly distributed in space and continuous in time. The purpose of data gridification is to provide an accurate initial field for ocean numerical prediction and ocean simulation. Commonly used data assimilation methods include two-dimensional optimal interpolation, three-dimensional variational, ensemble Kalman filter, etc. The biggest improvement of the optimal interpolation method (OI) is that the weights consider the statistical characteristics of the background field and observation errors, thus providing a more scientific and systematic analysis method. The analytical solution of three-dimensional variational (3D-VAR) is globally optimal, can handle the case where the observation operator is nonlinear, and can assimilate observation data from various sources. Based on 3D-VAR, 4D-VAR adds the assimilation of time variables, uses the complete dynamic model as a strong constraint within the time window to automatically adjust the model error in order to obtain a more accurate assimilation result, and incorporates the observation data in a certain time period into the assimilation system. The background field error covariance develops implicitly, and the error information propagates forward with the dynamic model, which is the main advantage of 4D-VAR. Since 4D-VAR needs to solve the adjoint model, and the cost function is usually solved by iterative calculations such as the steepest descent method, conjugate gradient method, and quasi-Newton iterative method, the computational amount is particularly large. The Kalman filter (KF) method is a new method for solving the recursive solution of the linear filtering problem of discrete data, which is a recursive data processing method under the assumptions that the system is linear, the noise is white, and the Gaussian strip condition is satisfied. With the gradual combination of the ensemble prediction idea and the research of the Kalman filter, the ensemble Kalman filter method (EnKF) is formed. It uses ensemble samples to estimate the background error covariance matrix. EnKF well solves the defects that the Kalman filter cannot calculate nonlinear problems and the background covariance matrix is difficult to predict. Moreover, this method also avoids the use of adjoint equations in the four-dimensional variational method.

[0003] However, in the prior art, in order to solve the density inversion problem caused by two-dimensional gridification, most people use three-dimensional variational (or four-dimensional) variational for overall gridification in three-dimensional space (or four-dimensional space-time). However, this solution requires a particularly large amount of calculation and is difficult to promote in practical applications. Secondly, the purpose of improving density inversion can also be achieved by calculating the vertical vector angle of the ocean observation profile (P-vector method), but the P-vector method requires a large amount of observation data. At present, the on-site observations in the global ocean are relatively few, and this method is ineffective in places with sparse observations.

[0004] In summary, when correcting the density inversion phenomenon of the ocean dataset, the existing technologies cannot provide an effective method for constructing a more accurate and higher-resolution ocean data grid product. Summary of the Invention

[0005] Based on this, in view of the above technical problems, it is necessary to provide a method for gridifying ocean data for correcting density inversion.

[0006] This specification adopts the following technical solutions:

[0007] This specification provides a method for gridifying ocean data for correcting density inversion, including:

[0008] Obtain multi-scale ocean data;

[0009] In a three-dimensional space, construct a three-dimensional regular grid, specify any grid point as an analysis grid point, calculate the three-dimensional correlation scale between the analysis grid point and the observation grid point, and determine the search radius for gridifying ocean data according to the change of the three-dimensional correlation scale with the distance between the analysis grid point and the target test point;

[0010] Within the determined search radius, calculate the background error correlation parameters between the analysis grid point and its adjacent observation grid points and the background error correlation parameters between two adjacent observation grid points through the three-dimensional correlation scale; and determine the square ratio of the observation error to the background error; and calculate the three-dimensional optimal weight between the analysis grid point and the observation grid point through the background error correlation parameters between the analysis grid point and its adjacent observation grid points, the background error correlation parameters between two adjacent observation grid points, and the square ratio of the observation error to the background error;

[0011] Obtain a gridified background field by performing arithmetic averaging on the multi-scale ocean data within a specified angle range of the analysis grid point;

[0012] Inverse interpolate the gridified background field to the observation grid points to obtain the observed values and background estimated values of the observation grid points;

[0013] Correct the analysis grid point through the observed value, background estimated value, three-dimensional optimal weight, and gridified background field of the observation grid point to obtain an analysis field; and perform iterative calculation on the multi-scale ocean data through the analysis field to correct the density inversion of the multi-scale ocean data and obtain a multi-scale analysis field.

[0014] Preferably, the obtaining of the multi-scale ocean data specifically includes:

[0015] Obtain ocean data through the China Real-time Data Center;

[0016] Perform quality control on the ocean data to obtain annual and monthly ocean data;

[0017] Among them, the quality control of the ocean data includes: water depth increasing test, temperature-salinity range test, freezing point test, spike signal detection, gradient test and triple variance test;

[0018] Classify the annual and monthly ocean data by month, quarter and year to obtain multi-scale ocean data including annual ocean data, quarterly ocean data and monthly ocean data.

[0019] Preferably, the calculation of the three-dimensional correlation scale between the analysis grid point and the observation grid point specifically includes:

[0020] Let the number of observation grid points of the analysis grid point in the three-dimensional space of horizontal and vertical directions be p, and the formula for calculating the three-dimensional correlation scale between the analysis grid point and the observation grid point is:

[0021]

[0022] In the formula, L' x 、L' y and L' z are the correlation scales between the horizontal, radial and vertical analysis grid points and the observation grid points respectively, L H and L V are correlation scale constants, are the gradients of the analysis variables in the meridional, zonal and vertical directions respectively, and E represents the gradient mean.

[0023] Preferably, the determination of the search radius for ocean data gridding specifically includes:

[0024] Determine the maximum value of the search radius based on the specified three-dimensional correlation scale;

[0025] When the three-dimensional correlation scale varies greatly with the distance between the analysis grid point and the target test point, divide the maximum value by a correlation scale parameter greater than 1 to reduce the search radius;

[0026] When the change of the three-dimensional correlation scale with the distance between the analysis grid point and the target test point is relatively small compared with something, divide the maximum value by a correlation scale parameter less than 1 to expand the search radius.

[0027] Preferably, the formula for calculating the background error correlation parameter between the analysis grid point and its adjacent observation grid point and the background error correlation parameter between two adjacent observation grid points is:

[0028]

[0029] In the formula, μ xy represents the background error correlation parameter between the analysis grid point and the observation grid point or between two observation grid points, r x 、r y 、r zis the meridional, zonal, and vertical distances between two points; L' x , L' y , L' z are the relevant scales in the meridional, zonal, and vertical directions, respectively.

[0030] Preferably, the formula for calculating the three-dimensional optimal weight of the observation grid point is:

[0031]

[0032] In the formula, w gj is the optimal weight between the analysis grid point and the observation grid point, μ jk is the background correlation error between two observation grid points, μ gk is the background error correlation parameter between the grid point and the observation grid point, η k is the square ratio of the observation error to the background error; k is the observation grid point.

[0033] Preferably, obtaining the analysis field specifically includes:

[0034] Inverse interpolating the gridded background field to the observation grid point to obtain the background estimate value, multiplying and accumulating the difference between each observation value and the background estimate value by the corresponding optimal weight w gj and adding it to the background value to obtain the analysis value. The formula is:

[0035]

[0036] In the formula, x ag is the analysis value at grid point g, x bg is the background value at grid point g, x oj is the j-th available observation value around grid point g, and H(x bj ) is the background estimate value of the j-th observation value.

[0037] Preferably, the square ratio of the observation error to the background error is used to reflect the proportion of the actual observation information of the ocean data and the information of the gridded background field.

[0038] Preferably, obtaining the multi-scale analysis field specifically includes:

[0039] Based on the gridded background field and annual ocean data, calculating and obtaining the annual ocean data analysis field using the analysis field;

[0040] Based on the annual ocean data analysis field and quarterly ocean data, calculating and obtaining the quarterly ocean data analysis field using the analysis field;

[0041] Based on the quarterly ocean data analysis field and monthly ocean data, calculating and obtaining the monthly ocean data analysis field using the analysis field;

[0042] The annual and monthly ocean data analysis fields are calculated based on the monthly ocean data analysis field and the annual and monthly ocean data utilization analysis field;

[0043] Among them, the relevant scales used in the calculation of the annual ocean data analysis field, the quarterly ocean data analysis field, and the monthly ocean data analysis field are all fixed values; the relevant scales used in the calculation of the annual and monthly data analysis field are obtained by calculating the annual ocean data analysis field;

[0044] The multi-scale analysis field includes: the annual ocean data analysis field, the quarterly ocean data analysis field, and the monthly ocean data analysis field.

[0045] The above at least one technical solution adopted in this specification can achieve the following beneficial effects:

[0046] In the ocean conditions where ocean observation data is sparse, the present invention first determines the grid search radius by calculating the three-dimensional correlation scales in the horizontal and vertical directions in the three-dimensional space, so that during the grid process, it is ensured that enough observation data is utilized to ensure that while fully extracting the effective information of the observation data, the calculation amount is not additionally increased. Secondly, based on determining the number of available observation grid points within the search radius of the analysis grid point, the background error related parameters are accurately calculated, and the optimal square ratio of the background error to the observation error is determined, extracting sufficient ocean data information to make the grid fusion structure closer to the real ocean structure, thus solving the problem of edge information distortion of ocean data; finally, by calculating the three-dimensional optimal weights of the analysis grid point and the observation grid point, the analysis grid points in the ocean data grid are effectively corrected, obtaining an analysis field containing ocean density information, and using the analysis field for iterative calculation to correct the multi-scale ocean data, obtaining an analysis field with more comprehensive and accurate ocean density information.

[0047] In summary, a method for gridifying ocean data with corrected density inversion provided by the present invention gridifies ocean data, calculates the three-dimensional optimal weights of the analysis grid point and the observation grid point, corrects the density inversion of the ocean data, and obtains a multi-scale analysis field containing more comprehensive ocean density information, providing an effective method for constructing a more accurate and higher-resolution ocean data grid product. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The drawings described herein are used to provide a further understanding of the present application, and constitute a part of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation of the present application. In the drawings:

[0049] Figure 1 is a schematic flow chart of a method for gridifying ocean data with corrected density inversion provided by this specification;

[0050] Figure 2 Schematic diagram of the gradient-dependent correction method based on a three-dimensional ellipsoid for a method of gridifying ocean data with corrected density inversion provided in this specification;

[0051] Figure 3 Correlation coefficient diagram of the difference between the density at different depths and the vertical density mean for a method of gridifying ocean data with corrected density inversion provided in this specification;

[0052] Figure 4 Relationship diagram of the search radius with the correlation coefficient and the optimal weight for a method of gridifying ocean data with corrected density inversion provided in this specification;

[0053] Figure 5 Comparison diagram of the gridified temperature profile results under different background error ratios for a method of gridifying ocean data with corrected density inversion provided in this specification;

[0054] Figure 6 Comparison diagram of the gridified salinity errors in the vortex region under different background error ratios for a method of gridifying ocean data with corrected density inversion provided in this specification;

[0055] Figure 7 Global ocean 300m layer density inversion correction comparison diagram for a method of gridifying ocean data with corrected density inversion provided in this specification;

[0056] Figure 8 Comparison station distribution diagram of the construction results in the mesoscale process region under different error ratios for a method of gridifying ocean data with corrected density inversion provided in this specification;

[0057] Figure 9 Cross-section comparison diagram of the results of different in-situ observation stations for a method of gridifying ocean data with corrected density inversion provided in this specification. Detailed implementation manners

[0058] To make the objectives, technical solutions, and advantages of this specification clearer, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of them. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of this application.

[0059] The following will detail the technical solutions provided by each embodiment of this application in conjunction with the drawings.

[0060] Figure 1 Flow schematic diagram of a method of gridifying ocean data with corrected density inversion in this specification, specifically including the following steps:

[0061] S101: Obtain multi-scale ocean data;

[0062] Obtain ocean data through the China Real-time Data Center;

[0063] Conduct quality control on the ocean data to obtain annual and monthly ocean data;

[0064] Among them, conducting quality control on the ocean data includes: water depth increasing test, temperature-salinity range test, freezing point test, spike signal detection, gradient test, and triple variance test;

[0065] Classify the annual and monthly ocean data by month, quarter, and year to obtain multi-scale ocean data.

[0066] S102: In three-dimensional space, construct a three-dimensional regular grid for the multi-scale ocean data, specify any grid point as the analysis grid point, calculate the three-dimensional correlation scale between the analysis grid point and the observation grid point, and determine the search radius for ocean data gridding according to the change of the three-dimensional correlation scale with the distance between the analysis grid point and the target test point, specifically including:

[0067] Assume that the number of observation grid points of the analysis grid point in the three-dimensional space of horizontal and vertical directions is p, and calculate the three-dimensional correlation scale between the analysis grid point and the observation grid point. The formula is:

[0068]

[0069] In the formula, L' x , L' y and L' z are the correlation scales between the horizontal, radial, and vertical analysis grid points and the observation grid points respectively, L H and L V are correlation scale constants, are the gradients of the analysis variables in the meridional, zonal, and vertical directions respectively, and E represents the gradient mean.

[0070] Determine the search radius for ocean data gridding, including:

[0071] Determine the maximum value of the search radius based on the specified three-dimensional correlation scale;

[0072] When the three-dimensional correlation scale changes greatly with the distance between the analysis grid point and the target test point, divide the maximum value by a correlation scale parameter greater than 1 to reduce the search radius;

[0073] When the change of the three-dimensional correlation scale with the distance between the analysis grid point and the target test point is relatively small compared with something, divide the maximum value by a correlation scale parameter less than 1 to expand the search radius.

[0074] Among them, the search radius is the local optimal horizontal spatial range during the gridification process. If the search radius is too large, the computational amount will double, and at the same time, some irrelevant false information will be introduced; if the search radius is too small, the available observation data around the grid points may be less, and the effective information in the observation data cannot be fully extracted. If the search radius is too small, the available observation data around the grid points may be less, and the effective information in the observation data cannot be fully extracted. Therefore, the determination of the search radius needs to be accurately tested in the application area according to time, location, and the number of observation data.

[0075] S103: Within the determined search radius, calculate and analyze the background error correlation parameters between the grid point and its adjacent observation grid points and the background error correlation parameters between two adjacent observation grid points through three-dimensional correlation scales; and determine the square ratio of the observation error to the background error; and calculate and analyze the three-dimensional optimal weights of the grid point and the observation grid point by analyzing the background error correlation parameters between the grid point and its adjacent observation grid points, the background error correlation parameters between two adjacent observation grid points, and the square ratio of the observation error to the background error, specifically including:

[0076] Calculate and analyze the background error correlation parameters between the grid point and its adjacent observation grid points and the background error correlation parameters between two adjacent observation grid points. The formula is:

[0077]

[0078] In the formula, μ xy represents the background error correlation parameter between the analysis grid point and the observation grid point or between two observation grid points, and r x , r y , r z are the zonal, meridional, and vertical distances between two points; L' x , L' y , L' z are the zonal, meridional, and vertical correlation scales respectively.

[0079] Calculate the three-dimensional optimal weight of the observation grid point. The formula is:

[0080]

[0081] In the formula, w gj is the optimal weight between the analysis grid point and the observation grid point, μ jk is the background correlation error between two observation grid points, μ gk is the background error correlation parameter between the grid point and the observation grid point, η k is the square ratio of the observation error to the background error; k is the observation grid point.

[0082] S104: Obtain the gridded background field by performing arithmetic averaging on multi-scale ocean data within a specified angle range of the analysis grid points.

[0083] S105: Inverse interpolate the gridded background field to the observation grid points to obtain the observed values and background estimated values of the observation grid points.

[0084] Among them, the square ratio of the error to the background error, this parameter plays an important role in the optimal interpolation objective analysis system, and it reflects the proportion of the actual observed information and background field information in the calculation result; the smaller the proportion of the observation error, the greater the weight of the data source in the gridded result; the greater the weight, the more fully the corresponding information is extracted. As long as the analysis grid points using this data are involved, theoretically, the corresponding gridded fusion structure is closer to the real ocean structure. However, in the area of meso-scale and small-scale dynamic processes, the horizontal and vertical gradients of ocean elements change very violently. Over-extracting the observed information in some areas (such as the vortex center) will cause the edge information to be distorted.

[0085] S106: Correct the analysis grid points through the observed values, background estimated values, three-dimensional optimal weights, and gridded background field of the observation grid points to obtain the analysis field; and perform iterative calculations on the multi-scale ocean data through the analysis field to perform density inversion correction on the multi-scale ocean data to obtain the multi-scale analysis field, specifically including:

[0086] Inverse interpolate the gridded background field to the observation grid points to obtain the background estimated values, multiply and accumulate the difference between each observed value and the background estimated value with the corresponding optimal weight w gj and add it to the background value to obtain the analysis value. The formula is:

[0087]

[0088] In the formula, x ag is the analysis value at grid point g, x bg is the background value at grid point g, x oj is the j-th available observed value around grid point g, and H(x bj ) is the background estimated value of the j-th observed value.

[0089] Based on the gridded background field and annual ocean data, use the analysis field to calculate and obtain the annual ocean data analysis field;

[0090] Based on the annual ocean data analysis field and quarterly ocean data, use the analysis field to calculate and obtain the quarterly ocean data analysis field;

[0091] Based on the quarterly ocean data analysis field and monthly ocean data, use the analysis field to calculate and obtain the monthly ocean data analysis field;

[0092] The annual and monthly ocean data analysis fields are calculated based on the monthly ocean data analysis field and the annual and monthly ocean data utilization analysis field;

[0093] Among them, the relevant scales used in the calculation of the annual ocean data analysis field, quarterly ocean data analysis field, and monthly ocean data analysis field are all fixed values; the relevant scales used in the calculation of the annual and monthly data analysis field are obtained by calculating the annual ocean data analysis field;

[0094] The multi-scale analysis field includes: annual ocean data analysis field, quarterly ocean data analysis field, monthly ocean data analysis field, and annual and monthly ocean data analysis field.

[0095] In this embodiment, taking the Argo temperature and salinity profile ocean data obtained from the China Argo Real-time Data Center as an example, through the above method steps, the density inversion correction of ocean data is realized, including:

[0096] Conduct water depth increasing test, temperature and salinity range test, freezing point test, spike (burr) signal detection, gradient test, and triple variance test on the annual and monthly ocean data;

[0097] Classify and summarize the multi-year data according to months, seasons (March, April, and May are spring, June, July, and August are summer, and so on), and multi-years (all data).

[0098] Construct an optimal interpolation scheme based on the three-dimensional ellipsoid related scale. For the scheme principle, see Figure 2 , in the horizontal and vertical three-dimensional space, calculate the observation data related to the analysis grid point ( Figure 4 ). That is, based on the number of available observations within the effective search radius around the determined analysis grid point, in the horizontal direction, according to the relevant scale between the analysis grid point and the observation grid point, calculate the optimal weight of each observation grid point; in the vertical direction, according to the analysis water layer and the vertical relevant scale, calculate the optimal weights of the available observation grid points in the upper and lower water layers, and then, through formulas (1)-(4), correct the analysis grid point with the observation true value, specifically including:

[0099] In the three-dimensional space R 3 Calculate the three-dimensional relevant scale at the analysis grid point through step S102. According to the research area and the data volume of the fused data, analyze the variation of the horizontal and vertical correlation coefficients with distance to determine the search radius of the interpolation method, while ensuring the use of sufficient observation data without additional computational effort;

[0100] Among them, for the determination of the vertical relevant scale parameter, see Figure 3, the correlation coefficient diagram of the difference between density at different depths and the mean vertical density. It can be seen from the figure that with the increase of depth, the correlation coefficients of the 5, 10, 15, and 20m depth layers gradually decrease, and the correlation coefficient below the 150m depth layer does not meet the research requirements.

[0101] In addition, the determination of the search radius requires precise testing in the application area based on time, location, and the amount of observation data. Figure 4 , which is a relationship diagram between the search radius, correlation coefficient and optimal weight in the global ocean in June 2024. During this period, the global average search radius is about 500-1000km.

[0102] By analyzing the three-dimensional correlation scale at the grid points, the background error correlation parameters between the observation grid points and the grid points and the background error correlation parameters between the two observation grid points are obtained;

[0103] Through the experimental analysis results of ocean data, the square ratio of observation error to background error is determined; among them, this parameter plays an important role in the optimal interpolation objective analysis system, which reflects the proportion of actual observation information and background field information in the calculation results;

[0104] See also Figure 5 , is a comparison chart of gridded temperature profile results under different background error ratios. From left to right, the error ratios between the observed data and the background field are 1:2, 1:5, and 1:10 respectively;

[0105] See also Figure 6 , is a comparison of the gridded salinity errors in the eddy region under different background error ratios. From left to right, the error ratios between the observed data and the background field are 1:2, 1:5, and 1:10, respectively;

[0106] The temperature-salinity profile within 2° around each specified grid point is arithmetic averaged to obtain the gridded background field, which provides the initial field for the subsequent data fusion.

[0107] The three-dimensional optimal weights of the analysis grid points and the observation grid points are obtained through the background error correlation parameters between the observation grid points and the grid points, the background error correlation parameters between the two observation grid points, and the square ratio of the observation error to the background error.

[0108] The initial field is interpolated to the observation grid to obtain the background estimate, and the difference between the true value of each observation grid point and the background estimate is compared with the corresponding optimal weight w gj The analysis value is obtained by multiplying and accumulating, and then adding it to the background value. At this point, the optimal interpolation scheme based on the relevant scale of the three-dimensional ellipsoid is completed.

[0109] Based on the gridded background field and annual profile data, an optimal interpolation scheme based on three-dimensional ellipsoid-related scales is used to fuse the profile data within a certain range around each grid point into the initial field with the calculated distance weights to obtain the annual climatological temperature and salinity field, that is, the annual ocean data analysis field;

[0110] Iteratively calculate the seasonal climatological temperature and salinity field and the monthly climatological temperature and salinity field. That is, based on the annual climatological temperature and salinity field data and the seasonal Argo temperature and salinity profile ocean data, using the optimal interpolation scheme based on three-dimensional ellipsoid-related scales, calculate the seasonal climatological temperature and salinity field. The monthly climatological temperature and salinity field is calculated based on the seasonal climatological temperature and salinity field data and the monthly Argo temperature and salinity profile ocean data. It should be noted that the relevant scales calculated above are all fixed values obtained through calculation; the annual and monthly temperature and salinity fields are calculated based on the monthly climatological temperature and salinity field and the annual and monthly Argo temperature and salinity profile ocean data, where the relevant scales are obtained through the calculation of the annual climatological temperature and salinity field.

[0111] See Figure 7 , which is a comparison chart of the density inversion correction for the 300m layer of the global ocean. The upper figure is the ocean density map before correction, and the lower figure is the ocean density map after correction. The blue dots are the grid points with density inversion, and the white ones are the grid points without density inversion.

[0112] See Figure 8 , which is a comparison station distribution map of the construction results in the mesoscale process area. The temperature and salinity profiles constructed by the optimal interpolation scheme based on three-dimensional ellipsoid-related scales are closer to the actual observations compared to the currently commonly used reanalysis products (HYCOM in the United States, GLORY in Europe, and the domestic operational products Modas and ISOP), and can retain more small and medium-scale ocean features;

[0113] See Figure 9 , which is a cross-sectional comparison chart of the results of different in-situ observation stations. The blue solid line (OI) in the figure is the profile constructed using the three-dimensional ellipsoid gradient-dependent method, the red solid line is the actual observation or Argo profile, Modas and ISOP are the existing domestic operational products respectively, HYCOM is the US reanalysis product, and GLORY is the European reanalysis product.

[0114] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity in description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as within the scope described in this specification.

Claims

1. A method for gridding ocean data with corrected density inversion, characterized in that: The following steps are involved: Acquire multi-scale ocean data; In three-dimensional space, a three-dimensional regular grid is constructed for multi-scale ocean data, and any grid point is designated as an analysis grid point. The three-dimensional correlation scale between the analysis grid point and the observation grid point is calculated, and the search radius of the ocean data gridding is determined according to the change of the three-dimensional correlation scale with the distance between the analysis grid point and the target test point. Within the determined search radius, the background error correlation parameters between the analysis grid point and the adjacent observation grid point and the background error correlation parameters between two adjacent observation grid points are calculated by three-dimensional correlation scale; and the square ratio of the observation error to the background error is determined; and the three-dimensional optimal weights of the analysis grid point and the observation grid point are calculated by the background error correlation parameters between the analysis grid point and the adjacent observation grid point, the background error correlation parameters between two adjacent observation grid points and the square ratio of the observation error to the background error; The gridded background field is obtained by arithmetically averaging the multi-scale ocean data within the specified angle range of the analysis grid points; The gridded background field is inversely interpolated to the observation grid points to obtain the observation values ​​of the observation grid points and the background estimation values; The analysis grids are corrected by using the observed values ​​of the observation grids, the estimated background values, the three-dimensional optimal weights and the gridded background field to obtain the analysis field. The multi-scale ocean data are iteratively calculated through the analysis field to perform density inversion correction on the multi-scale ocean data to obtain the multi-scale analysis field.

2. A method for gridding ocean data with corrected density inversion as claimed in claim 1, characterized in that: The obtaining of multi-scale ocean data specifically includes: Obtain ocean data through China Real-time Data Center; Conduct quality control on ocean data and obtain ocean data year by year and month by month; The quality control of ocean data includes: water depth increment test, temperature and salinity range test, freezing point test, peak signal detection, gradient test and three-fold variance test; The annual and monthly ocean data are classified according to month, quarter and year to obtain multi-scale ocean data including annual ocean data, quarterly ocean data and monthly ocean data.

3. A method for gridding ocean data with corrected density inversion as claimed in claim 1, characterized in that: The three-dimensional correlation scale between the calculation analysis grid and the observation grid specifically includes: Assuming that the number of observation grid points in the horizontal and vertical three-dimensional space of the analysis grid is p, the three-dimensional correlation scale between the analysis grid and the observation grid is calculated as follows: Where L' x , L' y and L' z are the correlation scales between the horizontal, radial and vertical analysis grids and the observation grids, respectively, and L H and L V is the related scaling constant, are the gradients of the analysis variables in the longitudinal, latitudinal and vertical directions respectively, and E represents the mean gradient.

4. A method for gridding ocean data with corrected density inversion as claimed in claim 1, characterized in that: Determining the search radius of ocean data gridding specifically includes: Determine the maximum value of the search radius based on the specified three-dimensional correlation scale; When the three-dimensional correlation scale varies greatly with the distance between the analysis grid point and the target test point, the maximum value is divided by a correlation scale parameter greater than 1 to reduce the search radius; When the three-dimensional correlation scale varies with the distance between the analysis grid point and the target test point and is relatively small, the maximum value is divided by a correlation scale parameter less than 1 to expand the search radius.

5. A method for gridding ocean data with corrected density inversion as claimed in claim 1, characterized in that: The calculation and analysis of the background error related parameters of the grid point and the adjacent observation grid point and the background error related parameters of the two adjacent observation grid points are as follows: In the formula, μ xy represents the background error related parameter between the analysis grid point and the observation grid point or between two observation grid points, r x 、r y 、r z is the distance between two points in the longitudinal, latitudinal and vertical directions; L' x , L' y , L' z They are the relevant scales in the longitude, latitude and vertical directions respectively.

6. A method for gridding ocean data with corrected density inversion as claimed in claim 1, characterized in that: The formula for calculating the three-dimensional optimal weight of the observation grid is: In the formula, w gj is the optimal weight between the analysis grid and the observation grid, μ jk is the background correlation error between two observation grid points, μ gk is the background error related parameter between the grid point and the observation grid point, η k is the square ratio of the observation error to the background error; k is the observation grid point.

7. A method for gridding ocean data with corrected density inversion as claimed in claim 1, characterized in that: The obtaining of the analysis field specifically comprises: The gridded background field is interpolated back to the observation grid to obtain the background estimate, and the difference between each observation value and the background estimate is compared with the corresponding optimal weight w gj Multiply and accumulate, and add it to the background value to get the analysis value. The formula is: In the formula, x ag is the analytical value at the grid point g, x bg is the background value at the grid point g, x oj is the jth observation value available around the grid point g, H(x bj ) is the background estimate of the jth observation.

8. A method for gridding ocean data with corrected density inversion as claimed in claim 1 or 4, characterized in that: The square ratio of the observation error to the background error is used to reflect the proportion of the actual observation information of the ocean data to the gridded background field information.

9. A method for gridding ocean data with corrected density inversion as claimed in claim 1, characterized in that: The acquisition of the multi-scale analysis field specifically includes: Based on the gridded background field and annual ocean data, the annual ocean data analysis field is obtained by using the analysis field calculation; Based on the annual ocean data analysis field and the quarterly ocean data, the quarterly ocean data analysis field is obtained by using the analysis field calculation; Based on the quarterly ocean data analysis field and the monthly ocean data, the monthly ocean data analysis field is obtained by using the analysis field calculation; Based on the monthly ocean data analysis field and the annual and monthly ocean data analysis field calculation, the annual and monthly ocean data analysis field is obtained; The relevant scales used in the calculation of the annual ocean data analysis field, the quarterly ocean data analysis field and the monthly ocean data analysis field are all fixed values; the relevant scales used in the calculation of the annual and monthly data analysis fields are obtained by calculating the annual ocean data analysis field; The multi-scale analysis field includes: an annual ocean data analysis field, a quarterly ocean data analysis field, a monthly ocean data analysis field, and a yearly and monthly ocean data analysis field.

Images