A method for spatiotemporal prediction of sea surface temperature

Abstract

The application discloses a shallow seawater temperature space-time prediction method, which is realized through the following steps: 1, collecting shallow seawater temperature environment data; 2, establishing a seawater surface temperature time sequence model; 3, establishing a shallow seawater temperature vertical structure model; 4, establishing a shallow seawater temperature vertical structure time-varying model; and 5, establishing a shallow seawater temperature space-time prediction model. Through the above steps, the time dynamic of the shallow seawater temperature is mapped to the parameters of the time-varying function for time scale prediction, and the effect of the shallow seawater temperature space-time prediction is further realized through the Kriging interpolation method. The practical problem that the existing method cannot accurately describe the time dynamic and regional difference of the shallow ocean temperature is solved, and the function of the space-time description of the large-range synchronous ocean temperature data is realized.

Description

Technical Field

[0001] This invention relates to a spatiotemporal prediction method for shallow seawater temperature, specifically a spatiotemporal prediction method for seawater temperature based on the theory of seawater temperature field stratification, considering temporal dynamics, error randomness, and regional differences. It is a spatiotemporal prediction method for seawater temperature based on a time-varying function model, Kriging interpolation, and a seawater temperature structure model. It holistically establishes a correlation model between seawater temperature observation data with different temporal and regional information and the structural characteristics of shallow seawater temperature, providing a quantitative model method for describing the distribution of shallow seawater temperature by relating it to time dimensions, latitude and longitude dimensions, and depth dimensions. It maps the temporal dynamics of shallow seawater temperature to the parameters of a time-varying function, facilitating time-scale prediction. Furthermore, it uses Kriging interpolation to conduct spatiotemporal prediction of shallow seawater temperature. This method is applicable to fields such as describing the spatiotemporal characteristics of seawater temperature and environmental prediction considering the physical and mathematical statistical relationships between surface and subsurface ocean information, and is an effective method for obtaining large-scale synchronous ocean temperature data. Background Technology

[0002] Seawater temperature is one of the most important environmental factors in ocean research, as it is a comprehensive result of ocean thermal, dynamic, and air-sea interactions. Seawater temperature prediction methods play a crucial role in climate change, ocean heat storage, ecological environment, national defense, and aquaculture. At the ocean surface, due to the mixing of atmospheric wind fields, seawater flows, and other natural environmental factors, a layer with approximately the same physicochemical properties forms, called the mixing layer. Below the mixing layer, temperature varies significantly with depth, forming the thermocline. Below the thermocline, the temperature change in deeper seawater tends to be slower. Therefore, the shallow seawater temperature variations within the mixing layer and thermocline below 200m of ocean surface are significant, making it a hot research topic in seawater temperature prediction. Traditional ocean data observation methods cannot obtain large-area, synchronous seawater temperature data. While remote sensing overcomes some of the shortcomings of traditional methods, its observations are limited to the ocean surface and cannot obtain vertical seawater temperature distribution information below the surface. Establishing a model of the relationship between ocean surface and subsurface seawater temperature would have significant practical implications for the study of ocean thermal structure, hydrodynamic processes, water masses, and other ocean phenomena.

[0003] Existing methods for studying the vertical structure of seawater temperature are mainly based on piecewise fitting, which involves dividing the ocean into 2 to 3 layers from top to bottom according to the characteristics of each layer and fitting them with different equations. The corresponding seawater stratification methods mainly include the temperature difference method, the gradient method, and the curvature method. In existing studies, researchers mostly focus on the fitting equations and fitting coefficients of each layer, while ignoring the impact of the temporal dynamics and regional differences of the ocean on stratification. For example, the upper and lower boundaries of the thermocline commonly found in the same ocean can vary significantly in different seasons, making it difficult to directly fit the vertical structure of seawater temperature using traditional piecewise fitting methods.

[0004] Based on this, the present invention combines seawater temperature observation data with different time and geographical information, as well as the structural characteristics of shallow seawater temperature and time-varying function models, to establish a quantitative model method for describing the distribution of shallow seawater temperature by establishing correlation models between time dimension, latitude and longitude dimension, depth dimension and seawater temperature, thereby achieving effective prediction of large-scale synchronous ocean temperature data.

[0005] The aforementioned "temperature difference method" refers to calculating the difference between the temperature of each data point in the seawater temperature field and the seawater surface temperature, and determining the boundary points of each layer of the ocean structure based on the temperature difference.

[0006] The "gradient method" refers to calculating the vertical gradient of each data point in the seawater temperature field and determining the boundary points of each layer of ocean structure based on the vertical gradient.

[0007] The “curvature method” refers to calculating the curvature of the vertical temperature distribution curve and using the maximum and minimum values ​​of curvature to determine the boundary points of the ocean's various structural layers.

[0008] The "fitting equation" refers to the equation that, given discrete data points, establishes a data relationship (mathematical model) and finds a series of tiny straight line segments that connect these discrete data points into a smooth curve. The function or parametric equation of this curve is the fitting equation.

[0009] The “fitting coefficient” refers to the numerical factor in the monomial of the algebraic expression of the fitted equation. The coefficient is usually not 0 and should be a rational number. Summary of the Invention

[0010] (1) Purpose of the invention:

[0011] To address the complex temporal and spatial variations in shallow seawater temperature and the lack of accurate and reasonable methods for spatiotemporal prediction of seawater temperature, this paper proposes a spatiotemporal prediction method for shallow seawater temperature. This method is a seawater temperature time-space prediction method based on temperature field stratification, considering temporal dynamics, error randomness, and regional differences. It is a seawater temperature spatiotemporal prediction method based on time-varying function models, Kriging interpolation, and seawater temperature structure models. It establishes correlation models between time, latitude and longitude, depth, and seawater temperature for seawater temperature observation data with different time and regional information and shallow seawater temperature structure characteristics, and uses these models to describe the quantitative model method of shallow seawater temperature distribution.

[0012] The "time dimension" refers to using time as a measurement scale to describe and express seawater temperature;

[0013] The "latitude and longitude" mentioned above refers to the use of geographical latitude and longitude coordinates as a measurement scale to describe and express seawater temperature;

[0014] The term "depth dimension" refers to using ocean depth as a metric to describe and express seawater temperature.

[0015] (2) Technical solution:

[0016] The following basic settings are required for this invention:

[0017] Setting 1: When predicting the sea temperature field, the sea temperature is treated as a random field in the horizontal direction. According to the first law of geography, all values ​​in geographic space are interconnected, and values ​​that are close to each other have a stronger connection.

[0018] The term "random field" refers to the natural generalization of the concept of random process in the spatial domain. It is regarded as a system of random variables defined on a parameter set in a spatial domain, where each point in the parameter set corresponds to a random variable.

[0019] The so-called "first law of geography" refers to the statement proposed by geographer Waldo R. Tobler that "all values ​​in geographic space are interconnected, and values ​​that are closer together are more closely connected."

[0020] Setting 2: The random process of seawater temperature field is an intrinsically stationary process, that is, the mathematical expectation of the random field exists and is independent of location; for any two points in the random field, the covariance function is only a function of the Euclidean distance between the points.

[0021]

[0022]

[0023] In the formula: Z is the estimation point to be interpolated, Zi Let λ be the measured point of the i-th sample, m be the number of measured samples involved in the calculation, and λ be the distance from the i-th sample. i The weight coefficient for the i-th sample point;

[0024] The "covariance function" refers to a function used to describe the overall error between variables in a space in a stochastic process or random field;

[0025] The "Euclidean distance" mentioned refers to a commonly used definition of distance, which is the actual distance between two points in multidimensional space, or the natural length of a vector (i.e., the distance from that point to the origin). In two-dimensional and three-dimensional space, the Euclidean distance is simply the actual distance between two points.

[0026] The “weighting coefficient” refers to the different proportions assigned to each sample in order to show the importance of several samples in the total sample.

[0027] Setting 3: In sea areas with a depth of less than 200 meters, it is assumed that the seawater temperature does not change significantly with the seasons, changes slowly with depth, and there is only a small temperature difference in the horizontal direction caused by seawater flow, with the seawater temperature fluctuating around 10 degrees Celsius.

[0028] Setting 4: Sea surface temperature is mainly affected by solar radiation, exhibiting both periodic and random fluctuation characteristics, with the overall temperature varying periodically around the annual average.

[0029] Setting 5: Due to the mixing effect of natural environmental factors such as atmospheric wind field and seawater flow, the seawater in the mixing layer forms a region with approximately the same physical and chemical properties. Therefore, it is assumed that the seawater temperature in the mixing layer is equal everywhere, and the vertical structure model of seawater temperature is simplified.

[0030] The “mixed layer seawater” refers to the uppermost layer of ocean water that is constantly in an active turbulent process due to the direct influence of air-sea interaction.

[0031] The method proposed in this invention mainly includes establishing a quantitative model method for describing the distribution of shallow seawater temperature by establishing correlation models between seawater temperature observation data with different time and geographical information and shallow seawater temperature structure characteristics, using time dimension, latitude and longitude dimension, depth dimension and seawater temperature. This enables effective prediction of large-scale synchronous ocean temperature data.

[0032] Based on the above assumptions and ideas, this invention provides a spatiotemporal prediction method for shallow seawater temperature, namely, a spatiotemporal prediction method for seawater temperature considering temporal dynamics, error randomness, and regional differences based on the theory of seawater temperature field stratification, which is implemented through the following steps:

[0033] Step 1: Collect shallow seawater temperature environmental data

[0034] Sufficient seawater temperature data is fundamental to establishing seawater forecasting models and supports subsequent steps. Seawater temperature data can be obtained from two sources: monitoring data from ship voyages and data from publicly available environmental databases. During a ship's voyage, the ship's trajectory is recorded, and seawater environmental data is continuously collected using seawater monitoring workstations. Collection parameters include collection time, seawater temperature, and seawater depth. After collection, the seawater environmental data from the ship's voyage is organized, requiring each data point to include four indicators: seawater temperature, latitude and longitude, time (accurate to year, month, day or year, month), and depth. Currently, publicly available environmental databases include the World Ocean Database (WOD), the Array for Real-Time Geostrophic Oceanography (ARGO), and the World Data Center for Climate (WDCC). Each data point collected from these databases is required to include four indicators: seawater temperature, latitude and longitude, time (accurate to year, month, day or year, month), and depth.

[0035] The aforementioned "WOD, ARGO, WDCC" refer to three representative publicly available marine environmental observation datasets used for observing and assessing changes in the marine environment.

[0036] Step 2: Establish a time-series model of sea surface temperature

[0037] In the marine environment, sea surface temperature is mainly affected by solar radiation and varies dynamically, thus exhibiting obvious periodic fluctuations with seasonal variations. Furthermore, ocean currents and atmospheric flows also disturb sea surface temperature, giving it stochastic fluctuation characteristics. Based on these two characteristics, a time-series model of sea surface temperature can be established using trigonometric functions.

[0038]

[0039] In the formula: T S (t) represents the average sea surface temperature in month t, where T0 is a constant. Used to describe the seasonal variation characteristics of sea surface temperature, T1 is the annual fluctuation amplitude of sea surface temperature, and τ and The fluctuation period and phase are related to latitude and longitude; ε(t) is a random fluctuation term, which typically follows a mean of 0 and a variance of σ. 2 The normal distribution of ε(t), i.e., ε(t) ~ N(0,σ). 2 (t));

[0040] Since solar radiation is primarily influenced by the Earth's revolution around the sun, the seasonal variation cycle of sea surface temperature is 12 months, i.e., τ = 12. T0 represents the annual average value of sea surface temperature, obtained by calculating the annual average sea surface temperature; the annual fluctuation amplitude T1 and phase of sea surface temperature are also considered. It is related to the area of ​​direct sunlight and varies with latitude and longitude. It needs to be specifically fitted by the sea surface temperature data of a certain location. ε(t) can be estimated by removing the fluctuation term from the temperature data and can be ignored when the sea surface temperature confidence interval is not considered.

[0041] Step 3: Establish a vertical structural model of shallow seawater temperature

[0042] For shallow seawater, the vertical temperature profile mainly consists of a mixing layer and a thermocline (the mixing layer is the uppermost layer of the ocean, directly affected by air-sea interaction and constantly in active turbulent flow; the thermocline is a thin layer located about 100-200 meters below the sea surface, with significant temperature and density variations, forming a layer where the water temperature drops sharply between the upper thin warm water layer and the lower thick cold water layer). The mixing layer is approximately a steady-state structure, with the seawater temperature approximately equal to the sea surface temperature. Within the thermocline, the rate of temperature decrease above 200 meters decreases with increasing depth. In areas below 200 meters, assuming no significant seasonal variation in seawater temperature and a slow change with depth, only a small temperature difference exists horizontally due to seawater flow, with the seawater temperature fluctuating around 10 degrees Celsius. Therefore, the seawater temperature in the mixing layer can be equated to the surface temperature, and the seawater temperature within the thermocline can be expressed as an exponential function of the sea surface temperature, temperature gradient parameters, and depth.

[0043] T h =T S ,h<h th (4)

[0044] T h =(T S -10)·exp[-b·(hh th )]+10,h>h th (5)

[0045] In the formula: T h The temperature of the seawater at a depth of h is represented by h. th denoted by , where b represents the upper boundary of the seawater thermocline, and b represents the seawater temperature gradient parameter.

[0046] The shallow seawater temperature vertical structure model of equations (4)-(5) is applicable to a depth of 200m below sea level; in this model, the upper boundary h of the seawater thermocline is... thIt needs to be determined by judging the rate of temperature change; the method for determining the upper boundary of the seawater thermocline is the vertical gradient method, that is, the lowest temperature gradient in the vertical direction is 0.05 degrees Celsius / meter to determine the upper boundary of the thermocline.

[0047] Step 4: Establish a time-varying model of the vertical structure of shallow seawater temperature.

[0048] Based on the vertical structure model of shallow seawater temperature, a time-varying function (a time-varying function is a function of variables where the independent variable is a time variable and the dependent variable changes with time) is introduced to describe the time-varying characteristics of the vertical structure of shallow seawater temperature (time-varying characteristics refer to the characteristic that the dependent variable changes with time). This allows for the establishment of a time-varying model of the vertical structure of shallow seawater temperature, specifically:

[0049] For the shallow seawater temperature vertical structure model described using equations (4)-(5), a total of h th Since h and b are two undetermined parameters that change with time, a time-varying function similar to equation (3) is introduced to describe h. th The characteristics of b changing with the seasons; considering that the shallow seawater thermocline is very unevenly distributed globally and is greatly affected by changes in depth, season, and latitude, directly reflecting the sea surface heat balance; therefore, the upper boundary h of the seawater thermocline... th The time-varying model should be similar to that of surface seawater temperature:

[0050]

[0051] Where: h th (t) represents the upper boundary depth of the shallow seawater thermocline in month t, and h0 is a constant. Used to describe the seasonal variation characteristics of the upper boundary depth of the shallow seawater thermocline, h1 is the annual fluctuation amplitude of the upper boundary depth of the shallow seawater thermocline, and τ and The oscillation period and phase are the same as those of the surface seawater.

[0052] Since setting 3 assumes that seawater temperature below 200 meters above sea level is not affected by seasons, the parameter b of the shallow seawater temperature index model is mainly determined by the mixing layer temperature. Since the mixing layer temperature is similar to the seawater surface temperature, the model of how the parameter b of the shallow seawater temperature index model changes over time should also be similar to that of the surface seawater temperature model.

[0053]

[0054] In the formula: b(t) represents the parameters of the shallow seawater temperature index model for month t, and b0 is a constant. The parameters of the shallow seawater temperature index model are used to describe the seasonal variation characteristics, where b1 is the annual fluctuation amplitude of the parameters of the shallow seawater temperature index model, and τ and The oscillation period and phase are the same as those of the surface seawater.

[0055] The two constants h0 and b0 contained in equations (6) and (7) reflect the annual average values ​​of the upper boundary depth of the shallow seawater thermocline and the parameters of the shallow seawater temperature index model, respectively, and are obtained by calculating the average values ​​of the constants over 12 consecutive months; the two annual fluctuation amplitude parameters h1 and b1 can be obtained by fitting regression (fitting regression refers to the statistical analysis method for studying the relationship between one set of random variables and another set of variables, also known as multiple regression analysis);

[0056] Step 5: Establish a spatiotemporal predictive model for shallow seawater temperature

[0057] The time-varying model of the vertical structure of shallow seawater temperature describes the variation of vertical seawater temperature over time at a certain location. However, under natural marine environmental conditions, seawater exhibits not only continuous changes on a time scale but also different patterns of change on a spatial scale. Therefore, based on the time-varying model of the vertical structure of shallow seawater temperature, a horizontal structure model of seawater temperature needs to be established to obtain large-scale synchronous ocean temperature data. This invention uses a Kriging interpolation method (based on semi-variogram theory and structural analysis, Kriging interpolation is a method for unbiased optimal estimation of regionalized variables within a finite region, and is one of the main contents of geostatistics) to describe the horizontal structure of seawater temperature, specifically:

[0058] The parameter vector of the time-varying model of the vertical structure of shallow seawater temperature Time data is correlated with latitude and longitude, and Kriging interpolation calculations are performed (Kriging interpolation calculations refer to the estimation of variables in a certain region based on the Kriging interpolation method); the Kriging equations are solved to obtain the parameter vectors of the time-varying model of the vertical structure of shallow seawater temperature at different latitudes and longitudes. To enable spatiotemporal prediction of shallow seawater temperature;

[0059] The shallow seawater temperature spatiotemporal prediction model maps the temporal dynamics of shallow seawater temperature onto the parameters of a time-varying function, facilitating time-scale prediction (time-scale prediction refers to the description, analysis, and prediction of future conditions and trends in the time dimension using modern scientific and technological means and methods based on information, experience, and laws already known to humankind in the past and present). This model further enables the development of shallow seawater temperature spatiotemporal prediction (seawater temperature spatiotemporal prediction refers to the description, analysis, and prediction of future conditions and trends in the time and space dimensions based on information, experience, and laws already known to humankind in the past and present regarding seawater temperature). It is applicable to fields such as the description of seawater temperature spatiotemporal characteristics and environmental prediction considering the physical and mathematical statistical relationships between surface and subsurface information, and is an effective method for obtaining large-scale synchronous ocean temperature data.

[0060] Through the above steps, the temporal dynamics of shallow seawater temperature are mapped onto the parameters of a time-varying function for time-scale prediction. Furthermore, the spatiotemporal prediction of shallow seawater temperature is achieved through Kriging interpolation. This solves the practical problem that existing methods cannot accurately describe the temporal dynamics and regional differences of shallow ocean temperature, and realizes the function of spatiotemporal description of large-scale synchronous ocean temperature data.

[0061] (3) Advantages and effects: This invention is a spatiotemporal prediction method for shallow seawater temperature, namely, a spatiotemporal prediction method for seawater temperature based on the theory of seawater temperature field stratification, considering the time dynamics, error randomness, and regional differences. Its advantages are:

[0062] ①This invention provides a quantitative model method for describing the distribution of shallow seawater temperature by establishing a holistic correlation model between time, latitude and longitude, depth and seawater temperature.

[0063] ②This invention considers the physical and mathematical statistical relationships between ocean surface and subsurface information. Compared with traditional modeling methods, it makes the prediction of seawater temperature more accurate and reasonable by using time-varying functions and Kriging interpolation.

[0064] ③ The parameters measured in this invention are basic physical parameters of seawater. The testing method is simple, easy to calculate, and the data processing method has low complexity.

[0065] ④ This prediction method is scientific and reasonable, has good processability, and has broad application value. Attached Figure Description

[0066] Figure 1 The flowchart of the method described in this invention.

[0067] Figure 2A schematic diagram of the time series model of seawater surface temperature in the marine area in this invention case.

[0068] Figure 3 A schematic diagram of the time-varying vertical structure of shallow seawater temperature in this invention case.

[0069] Figure 4 A schematic diagram of the predicted shallow seawater temperature in the present invention. Detailed Implementation

[0070] The present invention will be further described in detail below with reference to examples;

[0071] This invention discloses a spatiotemporal prediction method for shallow seawater temperature, specifically a spatiotemporal prediction method for seawater temperature based on the theory of seawater temperature field stratification, considering temporal dynamics, error randomness, and regional differences. (See attached text.) Figure 1 As shown, this can be achieved through the following steps:

[0072] Step 1: Collect shallow seawater temperature environmental data

[0073] In this case study, the shallow seawater temperature environmental data came from the World Climate Data Center. The selected data were the monthly average seawater temperature from January 2016 to December 2020. The depth data ranged from 0m to 200m below sea level, with one data point every 10m. The specific latitude and longitude information was 25°N and 122°E. Some data are shown in Table 1.

[0074] Table 1. Seawater Temperature Data (degrees Celsius)

[0075]

[0076] Step 2: Establish a time-series model of sea surface temperature

[0077] A time series model of sea surface temperature in this sea area was established using equation (3). Sixty sets of data were selected from January 2016 to December 2020, with T0 representing the average sea surface temperature over the 60 months. The calculation results are as follows:

[0078]

[0079] Based on the least squares method, equation (3) was fitted using 60 months of sea surface temperature data to obtain parameters T1 and Its goodness of fit reached 0.97, indicating the correctness of the model; the parameter fitting results are as follows:

[0080] T1 = 3.97℃

[0081]

[0082] The final time series model of sea surface temperature is as follows:

[0083]

[0084] The time series model and raw data of sea surface temperature in the case study are as follows: Figure 2 As shown, this model can accurately describe the temporal variation of sea surface temperature;

[0085] Step 3: Establish a vertical structural model of shallow seawater temperature

[0086] A vertical structure model of shallow seawater temperature was established using seawater temperature data at different depths in January 2016. First, the temperature gradient within each depth range was calculated, and the results are shown in Table 2.

[0087] Table 2 Seawater temperature gradients at different depth ranges

[0088]

[0089] The upper boundary of the thermocline is determined by the standard of the lowest temperature gradient in the vertical direction of 0.05 degrees Celsius / meter. According to the data in Table 2, in January 2016, the upper boundary of the thermocline in this sea area was 100m below sea level, and the mixed layer range was 0-100m.

[0090] Therefore, seawater temperature data in the 100-200m depth range were used to fit equation (5), and the goodness of fit was 0.974, which is relatively high, proving the correctness of the model. The fitting result obtained is as follows:

[0091] b = 0.02104

[0092] The vertical structure model of shallow seawater temperature in this sea area in January 2016 can be expressed as follows:

[0093]

[0094] The comparison chart between equation (9) and the original data is shown below. Figure 3 As shown, this model can accurately describe the variation of seawater temperature with depth;

[0095] Step 4: Establish a time-varying model of the vertical structure of shallow seawater temperature.

[0096] The selected data consisted of 60 sets of data from January 2016 to December 2020. The parameter fitting process in step three was repeated to obtain the parameters of the shallow seawater temperature vertical structure model for 60 months. Based on the shallow seawater temperature vertical structure model, a time-varying function was introduced to describe the time-varying characteristics of the shallow seawater temperature vertical structure for 60 months, and a time-varying model of the shallow seawater temperature vertical structure was established. Based on the least squares method to fit the parameters in equations (6) and (7), the results are shown in Table 3.

[0097] Table 3. Fitting results of time-varying model parameters for the vertical structure of shallow seawater temperature.

[0098]

[0099] Based on the parameters in Table 3, the upper boundary h of the seawater thermocline is obtained. th Models that change over time:

[0100]

[0101] The parameter b of the shallow seawater temperature index model changes with time as follows:

[0102]

[0103] Equations (10) and (11) respectively describe the variation of the upper boundary depth of the shallow seawater thermocline and the shallow seawater temperature index with time in this sea area;

[0104] Step 5: Establish a spatiotemporal predictive model for shallow seawater temperature

[0105] The World Climate Data Center website continued to collect monthly average sea surface temperature data from January 2016 to December 2020 at seven points. The depth data ranged from 0m to 200m below sea level, with one data point every 10m. The specific latitude and longitude of these points were: (25°N, 123°E), (25°N, 124°E), (24°N, 122°E), (23°N, 122°E), (23°N, 123°E), (23°N, 124°E), and (24°N, 124°E). Steps two through four were repeated for these seven coordinate points. After obtaining the time-varying vertical structure model of shallow sea surface temperature at eight coordinate points, Kriging interpolation was performed based on the latitude and longitude information of the eight coordinate points to establish a spatiotemporal predictive model of shallow sea surface temperature in the nearby sea area. The parameter estimation results for the eight coordinate points are shown in Table 4.

[0106] Table 4. Fitting results of time-varying model parameters for the vertical structure of shallow seawater temperature in the vicinity of the case.

[0107]

[0108] The coordinate parameters of the point (24°N, 123°E) obtained by Kriging interpolation are:

[0109]

[0110] The accuracy of the model was verified by predicting the sea surface temperature (24°N, 123°E) in January 2016. The prediction results are as follows: Figure 4As shown;

[0111] The results show that the method of the present invention can be used to analyze shallow seawater temperature data, map the temporal dynamics of shallow seawater temperature onto the parameters of a time-varying function for time-scale prediction, and further use Kriging interpolation to predict the spatiotemporal nature of shallow seawater temperature, thus achieving the expected purpose.

[0112] In summary, this invention relates to a spatiotemporal prediction method for shallow seawater temperature, specifically a spatiotemporal prediction method for seawater temperature based on the theory of seawater temperature field stratification, considering temporal dynamics, error randomness, and regional differences. It is a spatiotemporal prediction method for seawater temperature based on a time-varying function model, Kriging interpolation, and a seawater temperature structure model. It holistically establishes a correlation model between seawater temperature observation data with different temporal and regional information, as well as the structural characteristics of shallow seawater temperature, to describe the quantitative model method of shallow seawater temperature distribution by relating time, latitude / longitude, and depth dimensions to seawater temperature. It maps the temporal dynamics of shallow seawater temperature to the parameters of a time-varying function, facilitating time-scale prediction, and further utilizes Kriging interpolation to conduct spatiotemporal prediction of shallow seawater temperature.

[0113] The specific steps of this method are as follows: 1. Collect shallow seawater temperature environmental data; 2. Establish a time series model of seawater surface temperature; 3. Establish a vertical structure model of shallow seawater temperature; 4. Establish a time-varying model of the vertical structure of shallow seawater temperature; 5. Establish a spatiotemporal prediction model of shallow seawater temperature.

[0114] This invention is applicable to fields such as the description of the spatiotemporal characteristics of seawater temperature and environmental prediction, which consider the physical and mathematical statistical relationships between ocean surface and subsurface information. It features simple testing methods, low model complexity, good data fitting effect, and low data processing algorithm complexity, and has broad application value.

Claims

1. A method for spatiotemporal prediction of shallow seawater temperature, characterized in that: The following basic settings need to be established: Setting 1: When predicting the sea temperature field, the sea temperature is treated as a random field in the horizontal direction. According to the first law of geography, all values ​​in geographic space are interconnected, and values ​​that are close to each other are more strongly connected. Setting 2: The random process of seawater temperature field is an intrinsically stationary process, that is, the mathematical expectation of the random field exists and is independent of location; for any two points in the random field, the covariance function is only a function of the Euclidean distance between the points. In the formula: Z is the estimation point to be interpolated, Z i Let λ be the measured point of the i-th sample, m be the number of measured samples involved in the calculation, and λ be the distance from the i-th sample. i The weight coefficient for the i-th sample point; Setting 3: In sea areas with a depth of less than 200 meters, the seawater temperature is assumed to not change significantly with the seasons, and to change slowly with depth. There is only a small temperature difference in the horizontal direction caused by seawater flow, and the seawater temperature fluctuates by 10 degrees Celsius. Setting 4: Sea surface temperature is affected by solar radiation and has both periodic and random fluctuation characteristics. The overall temperature varies periodically around the annual average value. Setting 5: Due to the mixing effect of natural environmental factors such as atmospheric wind field and seawater flow, the seawater in the mixing layer forms a region with the same physical and chemical properties. Therefore, it is assumed that the seawater temperature in the mixing layer is equal, and the vertical structure model of seawater temperature is simplified. Based on the above basic settings, the steps are as follows: Step 1: Collect shallow seawater temperature environmental data Sufficient seawater temperature data is fundamental to establishing seawater forecasting models and supports subsequent steps. Seawater temperature data is acquired through two channels: monitoring data from ship voyages and data from publicly available environmental databases. During ship voyages, the ship's trajectory is recorded, and seawater environmental data is continuously collected using seawater monitoring workstations. Collection parameters include collection time, seawater temperature, and seawater depth. After collection, the seawater environmental data from the ship's voyages is processed, requiring each data point to include four indicators: seawater temperature, latitude and longitude, time, and depth. The time indicator must be accurate to the year, month, and day, or year and month. Currently, publicly available environmental databases include the World Oceans Dataset, the Global Ocean Observing Network, and the World Climate Data Center. Each data point collected from these databases must include the four indicators: seawater temperature, latitude and longitude, time, and depth. Step 2: Establish a time-series model of sea surface temperature In the marine environment, sea surface temperature dynamically changes due to the influence of solar radiation, thus exhibiting a significant seasonal fluctuation. Furthermore, ocean currents and atmospheric flows also disturb sea surface temperature, giving it a stochastic fluctuation characteristic. Based on these two characteristics of sea surface temperature, a time-series model of sea surface temperature is established using trigonometric functions: In the formula: T S (t) represents the average sea surface temperature in month t, where T0 is a constant. Used to describe the seasonal variation characteristics of sea surface temperature, T1 is the annual fluctuation amplitude of sea surface temperature, and τ and The fluctuation period and phase are related to latitude and longitude; ε(t) is a random fluctuation term, which follows a mean of 0 and a variance of σ. 2 The normal distribution of ε(t), i.e., ε(t) ~ N(0,σ). 2 (t)); Because solar radiation is affected by the Earth's revolution around the sun, the seasonal variation period of sea surface temperature is 12 months, i.e., τ = 12. T0 represents the annual average value of sea surface temperature, obtained by calculating the annual average sea surface temperature; the annual fluctuation amplitude T1 and phase of sea surface temperature are also considered. It is related to the area of ​​direct sunlight and varies with latitude and longitude, and needs to be specifically fitted by the sea surface temperature data of a certain location; ε(t) is estimated by removing the fluctuation term from the temperature data, and can be ignored when the sea surface temperature confidence interval is not considered. Step 3: Establish a vertical structural model of shallow seawater temperature For shallow seawater, the vertical temperature profile consists of a mixing layer and a thermocline. The mixing layer is approximately a steady-state structure, with the seawater temperature equal to the surface temperature. Within the thermocline, the rate of temperature decrease above 200 meters decreases with increasing depth. In areas below 200 meters, the seawater temperature is assumed to remain relatively constant with the seasons, changing slowly with depth, and exhibiting only a small temperature difference in the horizontal direction caused by seawater flow, fluctuating by approximately 10 degrees Celsius. Therefore, the temperature in the mixing layer is equivalent to the surface temperature, and the temperature within the thermocline can be expressed as an exponential function of the surface temperature, temperature gradient parameters, and depth. T h =T S , h<h th (4) T h =(T S -10)·exp[-b·(h-h th )]+10, h>h th (5) In the formula: T h The temperature of the seawater at a depth of h is represented by h. th denoted by , where b represents the upper boundary of the seawater thermocline, and b represents the seawater temperature gradient parameter. The shallow seawater temperature vertical structure model of equations (4)-(5) is applicable to a depth of 200m below sea level; in this model, the upper boundary h of the seawater thermocline is... th It needs to be determined by judging the rate of temperature change; the method for determining the upper boundary of the seawater thermocline is the vertical gradient method, that is, the lowest temperature gradient in the vertical direction is 0.05 degrees Celsius / meter to determine the upper boundary of the thermocline. Step 4: Establish a time-varying model of the vertical structure of shallow seawater temperature. Based on the vertical structure model of shallow seawater temperature, a time-varying function is introduced to describe the time-varying characteristics of the vertical structure of shallow seawater temperature, thus establishing a time-varying model of the vertical structure of shallow seawater temperature, specifically as follows: For the shallow seawater temperature vertical structure model described using equations (4)-(5), a total of h th Since h and b are two undetermined parameters that vary with time, a time-varying function is introduced to describe h. th The seasonal variation of b; the upper boundary of the ocean thermocline h th The time-varying model should be similar to the surface seawater temperature: Where: h th (t) represents the upper boundary depth of the shallow seawater thermocline in month t, and h0 is a constant. Used to describe the seasonal variation characteristics of the upper boundary depth of the shallow seawater thermocline, h1 is the annual fluctuation amplitude of the upper boundary depth of the shallow seawater thermocline, and τ and The oscillation period and phase are the same as those of the surface seawater. Since the seawater temperature at depths below 200 meters below sea level, as defined in setting 3, is no longer affected by the seasons, the parameter b of the shallow seawater temperature index model is determined by the mixed layer temperature. Since the mixed layer temperature is similar to the seawater surface temperature, the model for how the parameter b of the shallow seawater temperature index model changes over time should also be the same as that for the surface seawater temperature. In the formula: b(t) represents the parameters of the shallow seawater temperature index model for month t, and b0 is a constant. The parameters of the shallow seawater temperature index model are used to describe the seasonal variation characteristics, where b1 is the annual fluctuation amplitude of the parameters of the shallow seawater temperature index model, and τ and The oscillation period and phase are the same as those of the surface seawater. The two constants h0 and b0 contained in equations (6) and (7) reflect the annual average values ​​of the upper boundary depth of the shallow seawater thermocline and the parameters of the shallow seawater temperature index model, respectively, and are obtained by calculating the average values ​​of the constants over 12 consecutive months; the two annual fluctuation amplitude parameters h1 and b1 are obtained by fitting regression. Step 5: Establish a spatiotemporal predictive model for shallow seawater temperature The time-varying model of the vertical structure of shallow seawater temperature describes the variation of vertical seawater temperature over time at a certain location. However, under natural marine environmental conditions, seawater exhibits not only continuous changes on a time scale but also different patterns of change on a spatial scale. Therefore, based on the time-varying model of the vertical structure of shallow seawater temperature, a horizontal structure model of seawater temperature needs to be established to obtain large-scale synchronous ocean temperature data. A Kriging interpolation method is used to describe the horizontal structure of seawater temperature, specifically as follows: The parameter vector of the time-varying model of the vertical structure of shallow seawater temperature Time data is correlated with latitude and longitude. Kriging interpolation calculations are performed, and the Kriging equations are solved to obtain the parameter vectors of the time-varying model of the vertical structure of shallow seawater temperature at different latitudes and longitudes. To enable the spatiotemporal prediction of shallow seawater temperature.

Images