Argo-based method for constructing a spatial grid model of ocean dissolved oxygen

Abstract

The present application relates to the technical field of marine information, and more particularly to a method for constructing a marine dissolved oxygen spatial grid model based on Argo, comprising the following steps: obtaining climate state monthly scale dissolved oxygen data based on Argo dissolved oxygen profile data of the same month in previous years; constructing a relationship model between dissolved oxygen concentration and target depth determined by longitude and latitude coordinates corresponding to the climate state month from the climate state monthly scale dissolved oxygen data, to obtain dissolved oxygen concentration at different target depths in multiple longitudes and latitudes; dividing each target depth plane into multiple spatial units using longitude and latitude coordinates, and obtaining corresponding dissolved oxygen concentration and the marine dissolved oxygen spatial grid model through interpolation calculation of dissolved oxygen concentration of different spatial units at the same target depth. The present application avoids interference caused by differences in dissolved oxygen concentration between different data points through spatial heterogeneity partitioning, and overcomes the defect of discrete and sparse data in the Argo system by obtaining continuous data through interpolation calculation and weight calculation containing Argo quality marks.

Description

Technical Field

[0001] This invention relates to the field of marine information technology, and in particular to a method for constructing a marine dissolved oxygen spatial grid model based on Argo. Background Technology

[0002] Dissolved oxygen (DOXY) is oxygen dissolved in water. It provides the necessary biochemical environment for marine life and is an essential substance for marine life activities. The concentration of dissolved oxygen in seawater is not only a major indicator for measuring seawater quality and assessing the marine ecological environment, and an important basis for marine scientific experiments and resource exploration, but also a necessary parameter for understanding marine biogeochemical processes, global climate change, and the marine carbon cycle.

[0003] Current global dissolved oxygen concentration spatial grid data primarily relies on ship-based measurements, anchored buoys, and underwater intelligent detection equipment, resulting in poor continuous data updates. Ship-based measurements are the most common method, involving water sampling via CTD (Conductivity to Difference) water samplers followed by chemical titration analysis of continuous (discrete) water column samples. Existing ship-based methods suffer from low sampling rates and insufficient spatiotemporal resolution. Furthermore, shipborne observations are costly in terms of time and money, and are susceptible to extreme weather conditions, leading to scarce data in harsh sea states and polar regions. Currently, ocean dissolved oxygen observation data remains relatively sparse and unevenly distributed spatiotemporally, significantly limiting our understanding of the global ocean dissolved oxygen distribution and its influence by physical processes. Therefore, there is an urgent market demand for dissolved oxygen observation and analysis data processing methods that cover a wider temporal and spatial range. Summary of the Invention

[0004] In view of the above analysis, the present invention aims to provide a solution to at least one of the existing technical problems.

[0005] The objective of this invention is mainly achieved through the following technical solutions:

[0006] This invention provides a method for constructing a spatial grid model of dissolved oxygen in the ocean based on Argo, comprising the following steps:

[0007] Climatic monthly dissolved oxygen data were obtained by screening Argo dissolved oxygen profile data for the same month in previous years;

[0008] A model relating dissolved oxygen concentration to target depth was constructed using climatological monthly dissolved oxygen data, based on the latitude and longitude coordinates of the corresponding climatological month. This model yielded dissolved oxygen concentrations at different target depths under multiple latitude and longitude coordinates.

[0009] The plane corresponding to each target depth is divided into multiple spatial units using latitude and longitude coordinates. The dissolved oxygen concentration of each spatial unit at different target depths is obtained by interpolation calculation of the dissolved oxygen concentration of different spatial units at the same target depth, so as to construct the ocean dissolved oxygen spatial grid model.

[0010] Preferably, obtaining climatological monthly dissolved oxygen data includes: acquiring Argo historical dissolved oxygen profile data; and filtering by month to obtain historical climatological monthly dissolved oxygen data (Doxy) for each month. M (h); where M is 1-12 and h is the depth.

[0011] Preferably, the Argo dissolved oxygen profile data is obtained by scanning the same latitude and longitude coordinates (LO). j LA j The dissolved oxygen concentration and depth sequences at designated observation points corresponding to different target depths are denoted as (Doxy). ij Depth ij ); where Doxy ij Depth represents the dissolved oxygen concentration at the i-th observation point under the j-th latitude and longitude coordinates. ij Let be the depth value of the i-th observation point under the j-th latitude and longitude coordinates;

[0012] The method of obtaining dissolved oxygen concentrations at different target depths under multiple latitude and longitude coordinates includes:

[0013] Doxygen concentration data at different depths with known dissolved oxygen concentrations determined by latitude and longitude coordinates from corresponding climatological monthly dissolved oxygen data were used to establish Doxygen concentration data. ij and Depth ij The fitting function is: Doxy = f(Depth);

[0014] Based on the aforementioned relationship model, latitude and longitude coordinates, and target depth (Depth) ij The dissolved oxygen concentration (Doxy) at different target depths under multiple latitude and longitude coordinates was obtained. ij .

[0015] Preferably, the establishment of Doxy ij and Depth ij The fitting functions include:

[0016] Select the observation point (Doxy) from the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates. ij Depth ij The sampling points (x3, y3) and (x4, y4) are adjacent in depth; where y3 and y4 represent different depths, and y3 < Depth. ij<y4, x3 and x4 represent the dissolved oxygen concentration at the corresponding depth;

[0017] From the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates, sampling points (x1, y1) and (x2, y2) that are adjacent to sampling point (x3, y3) in depth are selected; from the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates, sampling points (x5, y5) and (x6, y6) that are adjacent to sampling point (x4, y4) in depth are selected; where y1, y2, y5, and y6 represent different depths, and y1 < y2 < y3, y4 < y5 < y6; x1, x2, x5, and x6 represent the dissolved oxygen concentration at the corresponding depth;

[0018] Construct observation points (Doxy) according to the following formula. ij Depth ij The Akima fitting function for ) is Doxy = f(Depth):

[0019] f(x) = p0 + p1(x - x3) + p2(x - x3) 2 +p3(x-x3) 3 ;

[0020] p0 = y3;

[0021] p1 = t3;

[0022]

[0023]

[0024] Where t3 is the first derivative at (x3, y3) and t4 is the first derivative at (x4, y4).

[0025] Preferably, the step of interpolating the dissolved oxygen concentration in different spatial units at the same target depth includes:

[0026] A subset A of dissolved oxygen concentrations at depth h at multiple latitude and longitude coordinates is constructed from dissolved oxygen concentrations at different target depths. h (Doxy uj LO j LA j ), for A h (Doxy uj LO j LA j Cluster analysis was performed to spatially heterogeneously partition the data points into different types of regions based on dissolved oxygen concentration and latitude / longitude coordinates, resulting in data subset A. h Spatial heterogeneity partitioning levels of the j-th latitude and longitude coordinates and the u-th data point at depth h: Q1, Q2, ..., Q n, where n is the number of grade intervals;

[0027] Based on spatial heterogeneity partitioning levels and data subset A h The distance between the data point and the center of the spatial unit, from data subset A h (Doxy uj LO j LA j Selecting data feature points from the data sets constitutes a subset A of the data. kh t From data subset A kh t (Doxy yj LO j LA j Obtain the dissolved oxygen concentration f in the background field of each space unit. b kh Where k represents the sequence of the kth spatial unit, b represents the background, h represents the depth, and t represents the data feature point;

[0028] Based on data subset A kh t (Doxy yj LO j LA j The quality label qc of the j-th latitude and longitude coordinate point and the y-th data point in the equation yj The distance d from the j-th latitude and longitude coordinate point to the y-th data point is from the center of the spatial unit. yj The spatial heterogeneity partitioning level of the j-th latitude and longitude coordinate point and the y-th data point yields the data subset A. kh t (Doxy yj LO j LA j The weight w of the j-th latitude and longitude coordinate point and the y-th data point at depth h. hyj ;

[0029] Based on the weight w hyj and dissolved oxygen concentration f in the background field of the space unit b kh The dissolved oxygen concentration at depth h of the k-th spatial unit is interpolated to obtain the dissolved oxygen concentration of each spatial unit at different target depths.

[0030] Preferably, the filtered data feature points include:

[0031] Using the center of the kth spatial unit as the center, divide the data subset A h (Doxy uj LO j LA jThe data points in the table are numbered C1, C2, ..., C6 from nearest to farthest. x , x represents the index of the x-th data point; data subset A kh t From the numbers C1, C2, ..., C x Data subset A consists of data feature points with the same spatial heterogeneity at the partition level. kh t satisfy:

[0032] 1) Select a subset of data feature points that satisfy the condition that the number of data feature points is ≥3 as a candidate data subset;

[0033] 2) Compare the parameter x of all candidate data subsets, and select the candidate data subset with the minimum value of x as data subset A. kh t .

[0034] Preferably, the step of obtaining the dissolved oxygen concentration f in the background field of each spatial unit is... b kh Including: data subset A kh t The mean dissolved oxygen concentration Doxy- of the data points is used as the background dissolved oxygen concentration f of the spatial cell at depth h. b kh .

[0035] Preferably, the weight w hyj Methods for obtaining [the information] include:

[0036] QC yj Substituting into the following formula, we obtain the quality label weight w of the y-th data point at the j-th latitude and longitude coordinate point. qcyj :

[0037]

[0038] Based on the distance d between the j-th latitude and longitude coordinates and the y-th data point and the center of the spatial unit. yj The spatial identifier weight w of the y-th data point at the j-th latitude and longitude coordinates is obtained using the following formula. syj :

[0039]

[0040] Where N is a subset of data A kh t Number of feature points in the data;

[0041] Based on the quality identification weight w qcyj and spatial identifier weight w syj The weight w of the y-th data point at depth h is obtained using the following formula.hyj w hyj Satisfy the following formula:

[0042] W hyj =W syj ×W qcyj .

[0043] Preferably, the interpolation calculation of the dissolved oxygen concentration at depth h of the k-th spatial unit includes:

[0044] Δf is obtained by iteratively applying equation (1) below;

[0045] If Δf < 1.00 umol / kg, or the maximum number of iterations is greater than or equal to 10, the final interpolation result of the dissolved oxygen concentration of the kth spatial unit at depth h is obtained by equation (2);

[0046] If Δf ≥ 1.00 μmol / kg and the maximum number of iterations is less than 10, the final interpolation result of the dissolved oxygen concentration of the kth spatial unit at depth h output by equation (2) is used again as f. kh Perform step-by-step iterations, repeating equation (2) for interpolation until the iteration threshold of Δf < 1.00 umol / kg is met or the maximum number of iterations is greater than or equal to 10.

[0047] Equation (1) is:

[0048] (2) The equation is:

[0049] Among them, f kh f represents the final interpolated result of the dissolved oxygen concentration in the k-th spatial cell at depth h. b kh Doxy is the initial background field value of the dissolved oxygen concentration in the k-th spatial unit. hyj A represents the data subset A related to the k-th spatial unit. kh t For the y-th data point, w hyj Let be the weight coefficient of the y-th data feature point in the k-th spatial unit, and j be the latitude and longitude coordinate sequence of the y-th data feature point; N is the data subset A. kh t Number of feature points in the data.

[0050] Preferably, the fitting function model is a polynomial fitting model that includes reciprocal elements.

[0051] Preferably, the relationship model is the Akima fitting function.

[0052] Compared with the prior art, the present invention can achieve at least one of the following beneficial effects:

[0053] (1) This invention constructs a fitting function model of dissolved oxygen concentration-depth for a target depth with unknown dissolved oxygen concentration at specific latitude and longitude coordinates based on climatological monthly dissolved oxygen data. It simulates and calculates continuous data at different depths from discrete data at different depths provided by the existing Argo profile database, which greatly makes up for the lack of data sparsity at different depths in the existing Argo profile database. Specifically, Akima interpolation fitting is used to fit discrete data points to form curves while taking into account the effect of the reciprocal factor, resulting in a smoother and more natural curve compared to other spline function fitting. At the same time, it makes the fitting converge near the interpolation point, reducing the possibility of overfitting near the interpolation point.

[0054] (2) Compared with the prior art, the present invention uses latitude and longitude coordinates to divide the plane corresponding to each target depth into different spatial units, and uses the dissolved oxygen concentration at the target depth to interpolate the dissolved oxygen concentration of different spatial units at the same target depth to obtain the dissolved oxygen concentration of different spatial units at the target depth, thereby obtaining the dissolved oxygen concentration of different spatial units at different target depths, which greatly makes up for the lack of data sparsity in the existing Argo profile database at different latitude and longitude coordinates.

[0055] (3) The present invention adopts a self-powered active buoy sampling system represented by the Argo system, which has lower usage costs and more flexible sampling methods compared to traditional ship sampling methods. It can adapt to the task scenario of simultaneous large-scale sampling in different locations. Compared with traditional passive buoy sampling, active buoys can complete sampling of a larger area through active movement, with higher sampling efficiency and resistance to ocean current systems, and can complete efficient and accurate sampling of a predetermined area.

[0056] (4) The present invention uses spatial heterogeneity partitioning of dissolved oxygen data at different depths for each latitude and longitude coordinate, which avoids interference caused by selecting heterogeneous regions with large differences in dissolved oxygen concentration when performing interpolation calculations for spatial units, and is conducive to improving the accuracy of interpolation calculations.

[0057] (5) This invention uses the k-means clustering algorithm to perform spatial heterogeneity partitioning of dissolved oxygen data at different depths with latitude and longitude coordinates. It has the advantages of fast convergence speed and high efficiency. At the same time, the spatial latitude and longitude coordinates and dissolved oxygen concentration used in the k-means clustering analysis of this invention have good consistency with the parameters of the subsequent spatial unit interpolation calculation. Therefore, the k-means clustering analysis results and the spatial unit interpolation calculation have better consistency and matching degree.

[0058] (6) This invention introduces both distance-related spatial weights and Argo data quality-related weights, while taking into account spatial unit distance and Argo data quality factors. This avoids interference from both factors on the weights, which helps reduce errors in data correlation assessment and improves the accuracy of interpolation calculation.

[0059] (7) This invention uses interpolation of the dissolved oxygen concentration at data feature points near the center of the spatial unit and the background dissolved oxygen concentration of the spatial unit to approximate and find the minimum value, thereby realizing the convergent extreme value Δf of the discrete input variable (dissolved oxygen concentration at data feature points); and uses Δf as the initial dissolved oxygen concentration f of the background field. b kh The correction parameters improve the accuracy of dissolved oxygen concentration calculation for space units compared to existing technologies.

[0060] In this invention, the above-described technical solutions can be combined with each other to achieve more preferred combinations. Other features and advantages of this invention will be set forth in the following description, and some advantages may become apparent from the description or be learned by practicing the invention. The objects and other advantages of this invention can be realized and obtained through the embodiments described and the accompanying drawings, which are particularly pointed out. Attached Figure Description

[0061] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts.

[0062] Figure 1 This is a flowchart of a method for constructing a marine dissolved oxygen spatial grid model based on Argo, according to the present invention. Detailed Implementation

[0063] The preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which constitute a part of the present invention and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0064] Based on the technical problems existing in the prior art, this invention proposes a spatial grid model of ocean dissolved oxygen based on global marine biogeochemical buoy profile data.

[0065] Argo (Array for real-time geostrophic oceanography) is the only global three-dimensional real-time observation system for the upper ocean in the field of biogeochemistry. It provides 240,000 dissolved oxygen profile data worldwide, providing an important data foundation for understanding and analyzing the current status and changing trends of dissolved oxygen in the global ocean.

[0066] Argo employs a powered active buoy sampling system, operating in a sequence of "descent - drift at a preset depth - re-descent - ascent for measurement - surface positioning and data transmission." Within a single operational cycle, the Argo buoy can only measure dissolved oxygen at different depths for a given latitude and longitude location during the ascent phase. However, due to the limitations of Argo's sampling scheme design, the system inevitably suffers from sparse data across different latitude and longitude coordinates.

[0067] Meanwhile, because Argo buoys were designed to prevent surface pollutants from entering the sensor with seawater and reducing observation accuracy, approximately 80% of Argo buoys do not conduct observations in the 0-3 dbar range of the ocean surface, resulting in missing surface dissolved oxygen data. Furthermore, due to the highly complex observation process and marine environment of Argo buoys, the dissolved oxygen data observed in the vertical direction is discrete and irregular, and the Argo system also suffers from data sparsity in the vertical direction.

[0068] Establish an interpolation method for ocean dissolved oxygen based on profile spatial characteristics and quality indicators, and develop a global climatological monthly-scale ocean dissolved oxygen spatial grid product at standard observation layer depth, so as to provide a data foundation for carrying out global marine ecological health assessment and sustainable development management.

[0069] Based on the problems discovered by the inventors during the research process, this invention discloses a method for constructing a marine dissolved oxygen spatial grid model based on Argo, which specifically includes the following steps:

[0070] Step 1: Obtain climatological monthly dissolved oxygen data by filtering Argo dissolved oxygen profile data for the same month in previous years;

[0071] Specifically, the method for obtaining the climatological monthly dissolved oxygen data is as follows: access the Argo historical dissolved oxygen profile data, and filter the historical climatological monthly dissolved oxygen data for month M (Doxy). M (h), where M ranges from 1 to 12, and h represents the depth. Argo dissolved oxygen profile data provides dissolved oxygen data (Doxy(Y, M, h)) at different depths from January 1, 2010 to December 31, 2021; historical dissolved oxygen data for different months are obtained by filtering by month: Doxy1(h), ..., Doxy M (h); where Y is the year.

[0072] Step 2: Construct a model of the relationship between dissolved oxygen concentration and target depth based on the latitude and longitude coordinates of the corresponding climatological month using dissolved oxygen data, and obtain the dissolved oxygen concentration at different target depths under multiple latitude and longitude coordinates.

[0073] Specifically, based on the climatological monthly scale dissolved oxygen data (Doxy) for month M, M(h) Obtain latitude and longitude coordinates (LO) j LA j Climatic monthly dissolved oxygen data Doxy M (h, LO) j LA j ), by Doxy M (h, LO) j LA j Construct latitude and longitude coordinates (LO) j LA j ) Target depth h where dissolved oxygen concentration is unknown t A fitting function model of dissolved oxygen concentration-depth was used to obtain latitude and longitude coordinates (LO) through the fitting function model. j LA j ) target depth h t Dissolved oxygen concentration Doxy M (h t ); j represents Doxy M The sequence number of the latitude and longitude coordinates of the data points in (h) is used to obtain the dissolved oxygen concentration at different target depths under multiple latitude and longitude coordinates.

[0074] Step 3: Divide the plane corresponding to each target depth into multiple spatial units using latitude and longitude coordinates. By interpolating the dissolved oxygen concentration of different spatial units at the same target depth, obtain the dissolved oxygen concentration of each spatial unit at different target depths, so as to construct the ocean dissolved oxygen spatial grid model.

[0075] Specifically, the dissolved oxygen concentration at different target depths under multiple latitude and longitude coordinates is spatially heterogeneous and partitioned according to the dissolved oxygen concentration and latitude and longitude coordinates. Data feature points at the same depth near the center of the spatial unit are selected, and the dissolved oxygen concentration of the spatial unit is calculated based on the quality label of the data feature points and the distance difference between the data feature points and the center of the spatial unit.

[0076] Compared with existing technologies, this invention constructs a fitting function model of dissolved oxygen concentration-depth for a target depth with unknown dissolved oxygen concentration at specific latitude and longitude coordinates based on climatological monthly dissolved oxygen data. It simulates and calculates continuous data at different depths from discrete data at different depths provided by the existing Argo profile database, which greatly makes up for the lack of data sparsity at different depths in the existing Argo profile database.

[0077] Compared with existing technologies, this invention uses latitude and longitude coordinates to divide the plane corresponding to each target depth into different spatial units. The dissolved oxygen concentration at the target depth is used to interpolate the dissolved oxygen concentration of different spatial units at the same target depth to obtain the dissolved oxygen concentration of different spatial units at that target depth. This greatly makes up for the lack of data sparsity in existing Argo profile databases at different latitude and longitude coordinates.

[0078] Compared with existing technologies, this invention adopts a self-powered active buoy sampling system represented by the Argo system, which has lower operating costs and more flexible sampling methods compared with traditional ship sampling methods. It can adapt to mission scenarios of simultaneous large-scale sampling in different locations. Compared with traditional passive buoy sampling, active buoys can complete sampling over a larger area through active movement, with higher sampling efficiency and resistance to ocean current systems, and can complete efficient and accurate sampling of a predetermined area.

[0079] Specifically, Argo dissolved oxygen concentration profile data are located in the same latitude and longitude coordinate system (LOC). j LA j ) Observation points are set up corresponding to different target depths, and the ocean dissolved oxygen concentration and depth sequence at the observation points is denoted as: (Doxy ij Depth ij ); where Doxy ij Depth represents the dissolved oxygen concentration at the i-th observation point under the j-th latitude and longitude coordinates. ij The depth value is the i-th observation point at the j-th latitude and longitude coordinates; the method for obtaining the dissolved oxygen concentration at the target layer depth is as follows:

[0080] Step 2.1: Establish a Doxygenation database by combining dissolved oxygen concentration data from different depths with known dissolved oxygen concentrations from climatological monthly dissolved oxygen data. ij and Depth ij The fitting function is: Doxy = f(Depth); preferably, the fitting function model can be a polynomial fitting model that includes the reciprocal element; more preferably, the fitting function model can be the Akima fitting function; the Akima interpolation method is used to interpolate the dissolved oxygen concentration in the climatological monthly scale dissolved oxygen data.

[0081] Step 2.2: Input the target depth (Depth) of the i-th observation point at the j-th latitude and longitude coordinates into the fitting function in Step 2.1. ij Obtain the ocean dissolved oxygen concentration Doxy at the i-th observation point. ij .

[0082] The specific principle of Akima interpolation is as follows: a cubic polynomial curve with a first derivative is established based on known data points near the interpolation point.

[0083] Compared with existing technologies, this invention uses Akima interpolation to fit discrete data points into curves while taking into account the effect of the reciprocal factor, resulting in a smoother and more natural curve compared to other spline function fitting methods.

[0084] Specifically, step 2.1 describes the establishment of Doxy ij and Depth ij The method for fitting the function is as follows:

[0085] Step 2.1.1: Using the Akima math package in the analysis software, select the corresponding observation point from the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates. ij Depth ij The sampling points (x3, y3) and (x4, y4) are adjacent in depth; where y3 and y4 represent different depths, and y3 < Depth. ij <y4, x3 and x4 represent the dissolved oxygen concentration at the corresponding depth;

[0086] Step 2.1.2: Select sampling points (x1, y1) and (x2, y2) that are adjacent to sampling point (x3, y3) in depth from the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates; select sampling points (x5, y5) and (x6, y6) that are adjacent to sampling point (x4, y4) in depth from the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates; where y5 and y6 represent different depths, and y4 < y5 < y6; x5 and x6 represent the dissolved oxygen concentration at the corresponding depths;

[0087] Specifically, using the Akima math package in Python, observation points (Doxy) are constructed according to the following formula. ij Depth ij The Akima fitting function for ) is Doxy = f(Depth):

[0088] f(x) = p0 + p1(x - x3) + p2(x - x3) 2 +p3(x-x3) 3 .

[0089] p0 = y3;

[0090] p1 = t3;

[0091]

[0092]

[0093] Where t3 is the first derivative at (x3, y3) and t4 is the first derivative at (x4, y4).

[0094] Compared with the prior art, the present invention uses the reciprocal of the sampling points near the interpolation point as the coefficient of the polynomial fitting term to fit discrete data points into a curve. Akima interpolation takes into account the effect of the reciprocal of the factors, so that the fitting converges near the interpolation point and reduces the possibility of overfitting near the interpolation point.

[0095] Specifically, the spatial unit segmentation method described in step 3 is as follows:

[0096] By using dissolved oxygen concentrations at different target depths under multiple latitude and longitude coordinates, the dissolved oxygen concentration at each depth of a spatial cell is interpolated to obtain a dataset F(Doxy) of dissolved oxygen concentration at different depths of the spatial cell. k Depth k LO k LA k LO k+1 LA k+1 LO k+2 LA k+2 LO k+3 LA k+3 ), where k represents the sequence of the k-th spatial unit; Depth k Doxy represents the intended research depth of the k-th spatial unit; k Depth represents the k-th spatial unit. k Dissolved oxygen concentration at depth; LO k LO k+1 LO k+2 LO k+3 This represents the longitude coordinates of the vertices arranged clockwise, and LO k ≤LO k+3 <LO k+1 ≤LO k+2 ; among which LA k LA k+1 LA k+2 LA k+3 This represents the latitude coordinates of the vertices arranged clockwise, with North latitude defined as positive and South latitude as negative, satisfying LA. k ≥LA k+1 >LA k+2 ≥LA k+3 Construct the target depth h t Dissolved oxygen concentration in different spatial units.

[0097] Specifically, step 3 involves interpolating the dissolved oxygen concentration in different spatial units at the same target depth as follows:

[0098] Step 3.1: Construct a subset A of dissolved oxygen concentration data at depth h at multiple latitude and longitude coordinates using dissolved oxygen concentrations at different target depths. h (Doxy uj LO j LA j ), for A h (Doxy uj LO j LA j Cluster analysis was performed to spatially heterogeneously partition the data points into different types of regions based on dissolved oxygen concentration and latitude / longitude coordinates, resulting in data subset A. h Spatial heterogeneity partitioning levels of the j-th latitude and longitude coordinates and the u-th data point at depth h: Q1, Q2, ..., Q n Where n is the number of rank intervals. Specifically, K-means clustering analysis can be used. Specifically, spatial heterogeneous partitioning is performed based on dissolved oxygen concentration and latitude / longitude coordinates.

[0099] Step 3.2:

[0100] Based on spatial heterogeneity partitioning levels and data subset A h The distance between the data point and the center of the spatial unit, from data subset A h (Doxy uj LO j LA j ) Select data feature points to form a data subset A kh t (Doxy yj LO j LA j ), consisting of data subset A kh t (Doxy yj LO j LA j Obtain the dissolved oxygen concentration f in the background field of each space unit. b kh ; k represents the sequence of the k-th spatial unit, b represents the background, h represents the depth, t represents the data feature point, and y represents A. kh t The y-th data point is represented by the latitude and longitude coordinates. Specifically, spatial heterogeneous partitioning is performed based on dissolved oxygen concentration and latitude and longitude coordinates, and data feature points at the same depth near the center of the spatial unit are selected.

[0101] Step 3.3: Based on data subset A kh t (Doxy yj LO j LA jThe quality label qc of the j-th latitude and longitude coordinate point and the y-th data point in the equation yj The distance d from the j-th latitude and longitude coordinate point to the y-th data point is from the center of the spatial unit. yj The spatial heterogeneity partitioning level of the j-th latitude and longitude coordinate point and the y-th data point yields the data subset A. kh t (Doxy yj LO j LA j The weight w of the j-th latitude and longitude coordinate point and the y-th data point at depth h. hyj Specifically, the dissolved oxygen concentration of a spatial cell is calculated based on the quality identifier of the data feature points and the difference in distance between the data feature points and the center of the spatial cell, thus obtaining the weight w. hyj .

[0102] Step 3.4: Based on the weight w hyj and dissolved oxygen concentration f in the background field of the space unit b kh Interpolation calculations are performed on the dissolved oxygen concentration at depth h of the k-th spatial unit to obtain the dissolved oxygen concentration of each spatial unit at different target depths. Specifically, based on the weights w of multiple data feature points... hyj The dissolved oxygen concentration of a spatial unit is obtained by analyzing the dissolved oxygen concentration of multiple data feature points.

[0103] Compared with existing technologies, this invention performs spatial heterogeneity partitioning on dissolved oxygen data at different depths for each latitude and longitude coordinate, avoiding interference caused by selecting heterogeneous regions with large differences in dissolved oxygen concentration when performing interpolation calculations for spatial units, which helps to improve the accuracy of interpolation calculations.

[0104] Specifically, the spatial heterogeneity partitioning described in step 3.1 can be implemented using Python software:

[0105] Step 3.1.1:

[0106] After importing dissolved oxygen concentration data at different target depths under multiple latitude and longitude coordinates into common formats such as XML, Python's K-means clustering analysis package was used to select data points at depth h from the dissolved oxygen concentration data to construct a subset A. h (Doxy uj LO j LA j ), and utilize subset A h (Doxy uj LO j LA j Construct a spatial feature vector S containing latitude, longitude, and dissolved oxygen concentration. j S j =(LO jLA j Doxy 1j ...Doxy uj ) T ; j is the sequence of the j-th latitude and longitude coordinates; u = 1, ..., g; g is a subset A h The number of data points;

[0107] Step 3.1.2:

[0108] This scheme includes two categories: latitude and longitude coordinates and dissolved oxygen concentration. The number of categories z is set to 2; the iteration error threshold e min Set to 0.00001;

[0109] Step 3.1.3:

[0110] The number of iterations t is initially set to 0, and z initial cluster centers C are obtained. b (t) =S b b = 1, ..., z;

[0111] Step 3.1.4:

[0112] Using t as the variable, after t iterations, by C b (t) Get C b0 (t) When |S j ×C b0 (t) |<|S j ×C b (t) |When b = 1, ..., z; and b ≠ b0; j = 1, ..., u; take sample S j Assigned to the b0th cluster domain D b0 (t) At this point, all m samples are divided into k clustering domains D. b (t) In the given information, b = 1, ..., z;

[0113] Step 3.1.5:

[0114] After obtaining b0 in step 3.1.4, calculate the new cluster centers according to the following formula:

[0115] Where b = 1, 2, ..., z; N b Let b be the number of samples contained in the b-th cluster.

[0116] Step 3.1.6:

[0117] Using t as a variable, for C b (t+1)Iterate until |C b (t+1) -C b (t) |<e min If b = 1, 2, ..., z, then stop iterating; if the result of the t-th iteration is a clustering scheme with z categories, then proceed to the next step; otherwise, iterate for t = t + 1 and then go to step 4.1.4.

[0118] Step 3.1.7:

[0119] Compare z and u. If k ≥ u, the clustering ends; if k < u, iterate k = k + 1 and then proceed to step 3.1.3.

[0120] Step 3.1.8:

[0121] The clustering analysis results output in step 3.1.7 are used to construct a spatial heterogeneity analysis dataset Q or imported into a GIS database to draw a spatial heterogeneity analysis map of the data points in dataset A at depth h; the spatial heterogeneity analysis dataset Q is further divided into subsets A according to different dissolved oxygen concentrations. h The j-th latitude and longitude coordinates and the u-th data point are divided into different spatial heterogeneity partition levels at depth h: Q1, Q2, ..., Q n n is the number of grade intervals; Doxy1, Doxy2, ..., Doxy n They are Q1, Q2, ..., Q n The dissolved oxygen concentrations satisfy: Doxy1 ≤ Doxy2 … ≤ Doxy n .

[0122] Compared with existing technologies, this invention uses the k-means clustering algorithm to perform spatial heterogeneity partitioning of dissolved oxygen data at different depths with latitude and longitude coordinates. It has the advantages of fast convergence speed and high efficiency. At the same time, the spatial latitude and longitude coordinates and dissolved oxygen concentration used in the k-means clustering analysis of this invention have good consistency with the parameters of subsequent spatial unit interpolation calculation. Therefore, the k-means clustering analysis results and spatial unit interpolation calculation have better consistency and matching degree.

[0123] Specifically, the dissolved oxygen concentration f in the background field of the space unit mentioned in step 3.2 b kh The method for obtaining it is as follows:

[0124] Step 3.2.1: Using the center of the kth spatial unit as the center, divide the data subset A... h (Doxy uj LO j LA j The data points in the table are numbered C1, C2, ..., C6 from nearest to farthest.x , x represents the index of the x-th data point; data subset A kh t From the numbers C1, C2, ..., C x Data subset A consists of data feature points with the same spatial heterogeneity at the partition level. kh t satisfy:

[0125] 1) Select a subset of data feature points that satisfy the condition that the number of data feature points is ≥3 as a candidate data subset;

[0126] 2) Compare the parameter x of all candidate data subsets, and select the candidate data subset with the minimum value of x as data subset A. kh t .

[0127] Step 3.2.2: From data subset A kh t Calculate the mean dissolved oxygen concentration of the data points (Doxy) - The dissolved oxygen concentration f in the background field at depth h, which serves as a spatial unit. b kh .

[0128] Compared with existing technologies, this invention uses the average dissolved oxygen concentration of at least three data feature points of the same spatial heterogeneity partition level closest to the center of the spatial unit as the dissolved oxygen concentration f of the spatial unit background field. b kh This effectively avoids interference from data points that are spatially close but have significantly different dissolved oxygen concentrations, thus reducing prediction errors and improving prediction accuracy.

[0129] Specifically, the weight w mentioned in step 3.3 hyj The method for obtaining it is as follows:

[0130] Step 3.3.1: Argo dissolved oxygen profile data includes a built-in quality indicator (qc) for the dissolved oxygen profile data. yj It is divided into three levels: 1, 2, and 3, with a quality indicator weight w. qcyj Satisfy the following formula:

[0131]

[0132] Obtain the mass identifier qc of the y-th data point at the j-th latitude and longitude coordinate point from the Argo dissolved oxygen profile data. yj ; will qc yj Substituting into the above formula, we obtain the quality label weight w of the y-th data point at the j-th latitude and longitude coordinate point. qcyj ;

[0133] Step 3.3.2:

[0134] Based on the distance d between the j-th latitude and longitude coordinates and the y-th data point and the center of the spatial unit. yj Obtain the spatial identifier weight w of the y-th data point at the j-th latitude and longitude coordinates. syj Spatial weight w syj Satisfy the following formula:

[0135]

[0136] N is a subset of data A kh t Number of feature points in the data;

[0137] Step 3.3.3:

[0138] Based on spatial identifier weight w syj Quality Identifier Weight w qcyj Obtain the weight w of the y-th data point at depth h at the j-th latitude and longitude coordinates. hyj w hyj Satisfy the following formula:

[0139] W hyj =W syj ×W qcyj .

[0140] Compared with existing technologies, this invention introduces both distance-related spatial weights and quality label weights related to the quality of Argo data itself. It also takes into account the distance of spatial units and the quality of Argo data itself, thus avoiding interference from both factors on the weights. This helps to reduce errors in the assessment of data relevance and improve the accuracy of interpolation calculations.

[0141] Specifically, step 3.4, the method for interpolating the dissolved oxygen concentration at depth h for the k-th spatial unit, is as follows:

[0142] Δf is calculated iteratively using equation (1);

[0143] If Δf < 1.00 umol / kg, or the maximum number of iterations is greater than or equal to 10, the final interpolation result of the dissolved oxygen concentration of the kth spatial unit at depth h is output by equation (2);

[0144] If Δf ≥ 1.00 μmol / kg and the maximum number of iterations is less than 10, the final interpolation result of the dissolved oxygen concentration of the kth spatial unit at depth h output by equation (2) is used again as f. kh Perform step-by-step iterations, repeating equation (2) for interpolation until the iteration threshold of Δf < 1.00 umol / kg is met or the maximum number of iterations is greater than or equal to 10.

[0145]

[0146]

[0147] Among them, f kh f represents the final interpolated result of the dissolved oxygen concentration in the k-th spatial cell at depth h. b kh Doxy is the initial background field value of the dissolved oxygen concentration at the k-th grid point. hyj A represents the data subset A related to the k-th spatial unit. kh t The y-th data point, where w hyj Let be the weight coefficient of the y-th data feature point in the k-th spatial unit, where j is the latitude and longitude coordinate sequence of the y-th data feature point; N is the data subset A. kh t Number of feature points in the data.

[0148] Compared with existing technologies, this invention introduces both distance-related spatial weights and quality label weights related to the quality of Argo data itself. It also takes into account the distance of spatial units and the quality of Argo data itself, thus avoiding interference from both factors on the weights. This helps to reduce errors in the assessment of data relevance and improve the accuracy of interpolation calculations.

[0149] Compared with existing technologies, this invention approximates and finds the minimum value by interpolating the dissolved oxygen concentration of data feature points near the center of the spatial cell and the background dissolved oxygen concentration of the spatial cell, thereby realizing the convergent extreme value Δf of the discrete input variable (dissolved oxygen concentration of data feature points); and uses Δf as the initial dissolved oxygen concentration f of the background field. b kh The correction parameters improve the accuracy of dissolved oxygen concentration calculation for space units compared to existing technologies.

[0150] As an example, the final interpolation results at a depth of 10m in global sea areas in July were calculated using the above method, and the calculation results were substituted into GIS software to obtain the dissolved oxygen concentration distribution map at a depth of 10m in global sea areas in July, overcoming the defect of sparse and discontinuous distribution of Argo profile data in global sea areas.

[0151] In existing technologies, the dissolved oxygen concentration of the spatial cell background field, obtained by simply averaging the dissolved oxygen concentrations of data feature points, is directly used as the dissolved oxygen concentration of the spatial cell without considering the influence weights of different data feature points. Compared with existing technologies, this invention approximates and finds the minimum value by interpolating the dissolved oxygen concentrations of data feature points near the center of the spatial cell and the dissolved oxygen concentration of the spatial cell background, thereby achieving the convergent extreme value Δf of the discrete input variable (dissolved oxygen concentration of data feature points); and uses Δf as the initial dissolved oxygen concentration f of the background field. b khThe correction parameters improve the accuracy of dissolved oxygen concentration calculation for space units compared to existing technologies.

[0152] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for constructing a spatial grid model of dissolved oxygen in the ocean based on Argo, characterized in that, Includes the following steps: Climatic monthly dissolved oxygen data were obtained by screening Argo dissolved oxygen profile data for the same month in previous years; A model relating dissolved oxygen concentration to target depth was constructed using climatological monthly dissolved oxygen data, based on the latitude and longitude coordinates of the corresponding climatological month. This model yielded dissolved oxygen concentrations at different target depths under multiple latitude and longitude coordinates. The plane corresponding to each target depth is divided into multiple spatial units using latitude and longitude coordinates. The dissolved oxygen concentration of each spatial unit at different target depths is obtained by interpolation calculation of the dissolved oxygen concentration of different spatial units at the same target depth, so as to construct the ocean dissolved oxygen spatial grid model. The method of interpolating the dissolved oxygen concentration in different spatial units at the same target depth includes: A subset A of dissolved oxygen concentrations at depth h at multiple latitude and longitude coordinates is constructed from dissolved oxygen concentrations at different target depths. h (Doxy) uj LO j LA j ), for A h (Doxy) uj LO j LA j Cluster analysis was performed to spatially heterogeneously partition the data points into different types of regions based on dissolved oxygen concentration and latitude / longitude coordinates, resulting in data subset A. h Spatial heterogeneity partitioning levels of the j-th latitude and longitude coordinates and the u-th data point at depth h: Q1, Q2, ..., Q n； Where n is the number of grade intervals; Based on spatial heterogeneity partitioning levels and data subset A h The distance between the data point and the center of the spatial unit, from data subset A h (Doxy) uj LO j LA j Selecting data feature points from the data sets constitutes a subset A of the data. kh t (Doxy) yj LO j LA j ), consisting of data subset A kh t (Doxy) yj LO j LA j Obtain the dissolved oxygen concentration f in the background field of each space unit. b kh Where k represents the sequence of the k-th spatial unit, b represents the background, h represents the depth, t represents the data feature point, and y represents A. kh t The y-th data point in the dataset, where j represents the latitude and longitude coordinates; Based on data subset A kh t (Doxy) yj LO j LA j The quality label qc of the j-th latitude and longitude coordinate point and the y-th data point in the equation yj The distance d from the j-th latitude and longitude coordinate point to the y-th data point is from the center of the spatial unit. yj The spatial heterogeneity partitioning level of the j-th latitude and longitude coordinate point and the y-th data point yields the data subset A. kh t (Doxy) yj LO j LA j The weight w of the j-th latitude and longitude coordinate point and the y-th data point at depth h. hyj ; Based on the weight w hyj and dissolved oxygen concentration f in the background field of the space unit b kh The dissolved oxygen concentration at depth h of the k-th spatial unit is interpolated to obtain the dissolved oxygen concentration of each spatial unit at different target depths.

2. The method for constructing a marine dissolved oxygen spatial grid model based on Argo as described in claim 1, characterized in that, The process of obtaining climatological monthly dissolved oxygen data includes: acquiring historical dissolved oxygen profile data from Argo; and filtering the data by month to obtain historical climatological monthly dissolved oxygen data for each month (Doxy). M (h); where M is 1-12 and h is the depth.

3. The method for constructing a marine dissolved oxygen spatial grid model based on Argo as described in claim 1, characterized in that, The Argo dissolved oxygen profile data were obtained by using the same latitude and longitude coordinates (LO). j LA j The dissolved oxygen concentration and depth sequences at the observation points corresponding to different target depths are obtained and denoted as (Doxy). ij Depth ij ); Among them, Doxy ij Depth represents the dissolved oxygen concentration at the i-th observation point under the j-th latitude and longitude coordinates. ij Let be the depth value of the i-th observation point under the j-th latitude and longitude coordinates; The method of obtaining dissolved oxygen concentrations at different target depths under multiple latitude and longitude coordinates includes: Doxy is established by using dissolved oxygen concentration data at different depths with known dissolved oxygen concentrations and determined latitude and longitude coordinates from corresponding climatological monthly dissolved oxygen data. ij and Depth ij Fitting function: Doxy = f(Depth); Based on the aforementioned relationship model, latitude and longitude coordinates, and target depth (Depth) ij The dissolved oxygen concentration (Doxy) at different target depths under multiple latitude and longitude coordinates was obtained. ij .

4. The method for constructing a marine dissolved oxygen spatial grid model based on Argo according to claim 3, characterized in that, The establishment of Doxy ij and Depth ij The fitting functions include: Select the observation point (Doxy) from the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates. ij Depth ij The sampling points (x3, y3) and (x4, y4) are adjacent in depth; where y3 and y4 represent different depths, and y3 < Depth. ij <y4, x3 and x4 represent the dissolved oxygen concentration at the corresponding depth; From the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates, sampling points (x1, y1) and (x2, y2) that are adjacent to sampling point (x3, y3) in depth are selected; from the Argo dissolved oxygen concentration profile data with the same latitude and longitude coordinates, sampling points (x5, y5) and (x6, y6) that are adjacent to sampling point (x4, y4) in depth are selected; where y1, y2, y5, and y6 represent different depths, and y1 < y2 < y3, y4 < y5 < y6; x1, x2, x5, and x6 represent the dissolved oxygen concentration at the corresponding depth; Construct observation points (Doxy) according to the following formula. ij Depth ij The Akima fitting function Doxy = f(Depth) is: ； ； ； ； ； Where t3 is the first derivative at (x3, y3) and t4 is the first derivative at (x4, y4).

5. The method for constructing a marine dissolved oxygen spatial grid model based on Argo according to claim 1, characterized in that, The filtered data feature points include: Using the center of the kth spatial unit as the center, divide the data subset A h (Doxy) uj LO j LA j The data points in the table are numbered C1, C2, ..., C6 from nearest to farthest. x , x represents the index of the x-th data point; data subset A kh t From the numbers C1, C2, ..., C x Data points constitute a subset A of data points that share the same spatial heterogeneity at the partition level; kh t satisfy: 1) Select a subset of data feature points that satisfy the condition that the number of data feature points is ≥3 as a candidate data subset; 2) Compare the parameter x of all candidate data subsets, and select the candidate data subset with the minimum value of x as data subset A. kh t .

6. The method for constructing an Argo-based spatial grid model of dissolved oxygen in the ocean according to claim 5, characterized in that, The dissolved oxygen concentration f in the background field of each space unit is obtained. b kh Including: data subset A kh t Calculate the mean dissolved oxygen concentration of the data points (Doxy) - The dissolved oxygen concentration f in the background field at depth h, which serves as a spatial unit. b kh .

7. The method for constructing an Argo-based spatial grid model of dissolved oxygen in the ocean according to claim 5, characterized in that, The weight w hyj The methods for obtaining it include: Quality labeling (QC) yj Substituting into the following formula, we obtain the quality label weight w of the y-th data point at the j-th latitude and longitude coordinate point. qcyj : ； Based on the distance d between the j-th latitude and longitude coordinates and the y-th data point and the center of the spatial unit. yj The spatial identifier weight w of the y-th data point at the j-th latitude and longitude coordinates is obtained using the following formula. syj : ； Where N is a subset of data A kh t Number of feature points in the data; Based on the quality identification weight w qcyj and spatial identifier weight w syj The weight w of the y-th data point at depth h is obtained using the following formula. hyj w hyj Satisfy the following formula: 。 8. The method for constructing a marine dissolved oxygen spatial grid model based on Argo according to claim 1, characterized in that, The interpolation calculation of the dissolved oxygen concentration at depth h of the k-th spatial unit includes: Δf is obtained by iterating through equation (1) below; If Δf < 1.00umol / kg, or the maximum number of iterations is greater than or equal to 10, the final interpolation result of the dissolved oxygen concentration of the kth spatial unit at depth h is obtained by the following equation (2); If Δf ≥ 1.00 μmol / kg and the maximum number of iterations is less than 10, the final interpolation result of the dissolved oxygen concentration of the kth spatial unit at depth h, output by equation (2) below, is used again as f. kh Perform step-by-step iterations, repeating equation (2) for interpolation until the iteration threshold of Δf < 1.00umol / kg is met or the maximum number of iterations is greater than or equal to 10; (1) The formula is: ; (2) The formula is: ; Among them, f kh f represents the final interpolated result of the dissolved oxygen concentration in the k-th spatial cell at depth h. b kh Doxy is the initial background field value of the dissolved oxygen concentration in the k-th spatial unit. hyj A represents the data subset A related to the k-th spatial unit. kh t For the y-th data point, w hyj Let be the weight coefficient of the y-th data feature point in the k-th spatial unit, and j be the latitude and longitude coordinate sequence of the y-th data feature point; N is the data subset A. kh t Number of feature points in the data.

9. The method for constructing an Argo-based spatial grid model of dissolved oxygen in the ocean according to claim 1, characterized in that, The relationship model is a polynomial fitting model that includes reciprocal elements.

Images