A sea level height prediction method and system, and readable storage medium

Abstract

The present invention discloses a method and system for predicting sea level height, using a sea level height prediction model established for a target area to obtain sea level height data of each effective grid in the target area at a preset time, wherein the sea level height prediction model includes a first model item describing the change law of the sea level height trend component of each effective grid in the target area over time, a second model item describing the change law of the sea level height periodic component of each effective grid in the target area over time, and a third model item describing the change law of the sea level height residual component of each effective grid in the target area over time. The present invention realizes random dynamic prediction of sea level height through the established sea level height prediction model, which can reduce the amount of calculation and improve the prediction accuracy in predicting sea level height of a large spatial scale. The present invention also discloses a computer-readable storage medium.

Description

Technical Field

[0001] The present invention relates to the field of satellite data analysis and application, and in particular to a method and system for predicting sea level height. The present invention also relates to a computer-readable storage medium. Background Art

[0002] In 2018, the Intergovernmental Panel on Climate Change (IPCC) released the Special Report on Global Warming of 1.5°C, which elaborated on the impact and prospects of controlling global warming at 1.5°C. Among them, sea level rise is one of the factors that have a greater impact on humans. Sea level rise leads to changes in the average sea level, especially an increase in the occurrence of extreme sea level events, such as storm surges, floods, and increased intensity of tropical cyclones, which seriously affects the development of coastal cities.

[0003] In order to improve the accuracy of sea level change prediction, many scholars have carried out a series of research. At present, there are two main methods for predicting sea level height, namely model simulation prediction and empirical statistical prediction. However, the existing methods are complex in calculation and the accuracy of sea level prediction needs to be improved. Summary of the invention

[0004] The purpose of the present invention is to provide a sea level prediction method and system, which realizes the random dynamic prediction of sea level by establishing a sea level prediction model, and can reduce the amount of calculation and improve the prediction accuracy in predicting the sea level of a large spatial scale compared with the prior art. The present invention also provides a computer readable storage medium.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A sea level prediction method, using a sea level prediction model established for a target area to obtain sea level data of each valid grid in the target area at a preset time;

[0007] The sea level prediction model includes a first model item, a second model item and a third model item. The first model item describes the change law of the sea level trend component of each valid grid in the target area over time, the second model item describes the change law of the sea level periodic component of each valid grid in the target area over time, and the third model item describes the change law of the sea level residual component of each valid grid in the target area over time. The sea level trend component refers to the part of the sea level that shows a linear change trend over time, the sea level periodic component refers to the part of the sea level that shows a periodic change over time, and the sea level residual component refers to the part of the sea level other than the trend component and the periodic component.

[0008] Preferably, the first model term describes the change of the sea level height trend component of each effective grid in the target area over time by a linear function.

[0009] Preferably, the second model term describes the change of the periodic component of the sea level height of each effective grid in the target area over time by a sine function and / or a cosine function.

[0010] Preferably, constructing the second model item includes: performing spectral analysis on the part obtained by subtracting the trend component from the sea level height of each valid grid in the target area, selecting a main period from the obtained periods according to the amplitude values ​​corresponding to the obtained periods, and constructing the second model item with the sine function and / or cosine function represented by the main period.

[0011] Preferably, the third model term includes a model obtained by multiplying and accumulating a prediction matrix of the main space matrix time series of the sea level height residual component of each valid grid in the target area and an eigenvector matrix of the main space matrix.

[0012] Preferably, constructing the third model item includes:

[0013] Obtain the main space matrix of the portion of the sea level height of each effective grid in the target area minus the trend component and the periodic component and the corresponding time series;

[0014] According to the time series corresponding to the main space matrix, the eigenvector matrix of the main space matrix is ​​obtained;

[0015] A prediction matrix is ​​established for the time series corresponding to the main space matrix, and the prediction matrix corresponding to the main space matrix is ​​multiplied and added with the eigenvector matrix of the main space matrix to construct the third model item.

[0016] A sea level prediction system is used to execute the sea level prediction method described above.

[0017] Preferably, including:

[0018] An acquisition device, used to acquire the sea level height prediction model for the target area;

[0019] The prediction device is used to use the sea level prediction model to obtain the sea level data of each valid grid in the target area at a preset time.

[0020] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the sea level prediction method as described above are implemented.

[0021] It can be seen from the above technical scheme that a sea level prediction method and system provided by the present invention uses a sea level prediction model established for a target area to obtain sea level data of each effective grid in the target area at a preset time, wherein the sea level prediction model includes a first model item describing the change law of the sea level trend component of each effective grid in the target area over time, a second model item describing the change law of the sea level periodic component of each effective grid in the target area over time, and a third model item describing the change law of the sea level residual component of each effective grid in the target area over time, the sea level trend component refers to the part of the sea level that shows a linear change trend over time, the sea level periodic component refers to the part of the sea level that shows a periodic change over time, and the sea level residual component refers to the part of the sea level other than the trend component and the periodic component. The present invention realizes random dynamic prediction of sea level by establishing a sea level prediction model. Compared with the prior art, it can reduce the amount of calculation and improve the prediction accuracy in predicting the sea level of a large spatial scale.

[0022] The computer-readable storage medium provided by the present invention can achieve the above-mentioned beneficial effects. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0024] Figure 1 A schematic diagram of a sea level prediction model used in a sea level prediction method provided in an embodiment of the present invention;

[0025] Figure 2It is a schematic diagram of sea level height data at a geographical location (110°E, 5°N) and a first model term established for a trend component in a specific example;

[0026] Figure 3 is a schematic diagram of sea level height data at a geographical location (110°E, 5°N) and a second model term established for a periodic component in a specific example;

[0027] Figure 4 The remaining portion after removing the trend component and the periodic component from the sea level height data at the geographical location (110°E, 5°N) in a specific example;

[0028] Figure 5-1 to Figure 5-6 In order Figure 4 The first 6 principal component time series and predicted values ​​obtained from the data analysis are shown;

[0029] Figure 6-1 to Figure 6-6 In order Figure 4 The first six main modes obtained from the data analysis are shown;

[0030] Figure 7-1 The sea level height data at the geographical location (110°E, 5°N) in January 2018 measured by a satellite altimeter;

[0031] Figure 7-2 The sea level height data at the geographical location (110°E, 5°N) in January 2018 predicted using the method of an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0033] This embodiment provides a sea level prediction method, which uses a sea level prediction model established for a target area to obtain sea level data of each valid grid in the target area at a preset time. Figure 1 , Figure 1 A schematic diagram of a sea level prediction model used in a sea level prediction method provided in this embodiment. As shown in the figure, the sea level prediction model established for the target area includes a first model item 101, a second model item 102 and a third model item 103.

[0034] Among them, the first model item 101 describes the change law of the sea level height trend component of each valid grid in the target area over time, the second model item 102 describes the change law of the sea level height periodic component of each valid grid in the target area over time, and the third model item 103 describes the change law of the sea level height residual component of each valid grid in the target area over time. The sea level height trend component refers to the part of the sea level height that shows a linear change trend over time, the sea level height periodic component refers to the part of the sea level height that shows a periodic change over time, and the sea level height residual component refers to the part of the sea level height other than the trend component and the periodic component.

[0035] In the method of this embodiment, the sea level height is divided into three parts, including a part showing a linear change trend over time, a part showing a periodic change over time, and a part other than the trend component (i.e., the part showing a linear change trend over time) and the periodic component (i.e., the part showing a periodic change over time). Corresponding model items are established for each part of the sea level height to describe the change pattern of each part of the sea level height over time.

[0036] The target area refers to the geographical area where the sea level needs to be predicted. The target area is divided into grids, and the effective grids refer to the grids with sea level data. By using the sea level prediction model established for the target area, the sea level data of each effective grid in the target area at a preset time can be obtained. The sea level prediction method of this embodiment realizes the random dynamic prediction of the sea level through the established sea level prediction model. Compared with the prior art, it can reduce the amount of calculation in predicting the sea level of a large spatial scale and improve the prediction accuracy.

[0037] The sea level prediction method is described in detail below in conjunction with specific implementation methods.

[0038] Optionally, the first model item 101 may describe the change of the trend component of the sea level height of each effective grid in the target area over time by a linear function, that is, the trend component of the sea level height is considered to be the part of the sea level height that changes linearly over time.

[0039] Optionally, the first model item 101 can be constructed by the following method, and the first model item 101 can be obtained by performing a linear regression analysis on the sea level data of the effective grids in the target area. For the time series of the sea level data of each effective grid in the target area, a linear regression analysis is performed to obtain a linear regression model, which can be expressed as: y=a0+a1x+ε, y represents the dependent variable, x represents the independent variable, a0 represents the intercept, a1 represents the slope, a0 and a1 are model parameters, ε represents the error value, and represents a normally distributed error sequence with a mean of 0. The slope a0 represents the change trend of the time series of the sea level data.

[0040] Preferably, in order to accurately describe the relationship between the independent variable and the dependent variable, the least squares method can be used to estimate the model parameters, which is expressed as:

[0041]

[0042] Among them, y i represents the value of the dependent variable at time i, x i represents the value of the independent variable at time i, and n represents the length of the time series. Dependent variable y = [y1, y2, ..., y n ], independent variable x=[x1,x2,…,x n ].

[0043] Thus, the first model term of the sea level prediction model is constructed according to the obtained linear regression model. The first model term can be expressed as: T(t)=a0+a1t.

[0044] Optionally, the second model item 102 may describe the change of the periodic component of the sea level height of each effective grid in the target area over time by using a sine function and / or a cosine function.

[0045] Optionally, the second model item 102 can be constructed by the following method: spectral analysis can be performed on the data of the sea level height of each valid grid in the target area minus the trend component, and a main period can be selected from the obtained periods according to the amplitude corresponding to each obtained period, and the second model item 102 can be constructed with the sine function and / or cosine function represented by the main period.

[0046] Further, the second model item 102 may be constructed according to the following process, which mainly includes the following steps:

[0047] S20: performing frequency spectrum analysis on the portion of the sea level height of each effective grid in the target area minus the trend component to obtain a plurality of cycles.

[0048] The data obtained by subtracting the trend component T(t) from the original sea level height data SLH(t) is subjected to spectrum analysis. The spectrum analysis may be specifically Fourier spectrum analysis to obtain multiple cycles. An exemplary expression may be:

[0049]

[0050] Wherein, Y(t) represents the remaining part of the sea level height data after removing the trend component, Y(ω) represents the frequency spectrum obtained by performing Fourier transform of N points of Y(t), and ω represents the period value.

[0051] S21: Selecting a main period from the obtained periods according to the amplitudes corresponding to the obtained periods.

[0052] Optionally, according to the amplitudes corresponding to the obtained periods, the period with an amplitude greater than a preset value can be selected from the obtained periods as the main period. Exemplarily, the amplitudes corresponding to each period in the spectrum sequence Y(ω) are traversed, and the highest amplitude is expressed as A max , you can choose higher than 0.1A max The period corresponding to the amplitude of is the main period.

[0053] S22: Construct the second model term using a sine function or / and a cosine function represented by a main period.

[0054] Optionally, the obtained main cycle can be used to construct a second model item 102 of the sea level prediction model based on the harmonic analysis model, that is, a model item describing the change law of the sea level periodic component of each valid grid in the target area over time is constructed. Exemplarily, assuming that the sequence contains K main cycles, the second model item 102 can be expressed by the following formula:

[0055]

[0056]

[0057] Among them A k and B k Related to the main cycle:

[0058]

[0059] Optionally, the third model item 103 may include a model obtained by multiplying and accumulating a prediction matrix of the main space matrix time series of the sea level height residual component of each valid grid in the target area and an eigenvector matrix of the main space matrix.

[0060] Optionally, the third model item 103 may be constructed by the following method, which mainly includes the following steps:

[0061] S30: Obtaining the main space matrix of the portion of the sea level height of each effective grid in the target area minus the trend component and the periodic component and the corresponding time series.

[0062] Optionally, an empirical orthogonal function analysis can be performed on the portion of the sea level height SLH(t) of each valid grid in the target area minus the trend component T(t) and the periodic component P(t), that is, the residual data portion of the sea level height of each valid grid in the target area, to calculate each spatial matrix and the time series corresponding to each spatial matrix, that is, to obtain each EOF mode and the time series corresponding to each EOF mode.

[0063] The main mode is selected from each EOF mode obtained. Optionally, the main mode can be selected from each EOF mode according to the cumulative variance. Exemplarily, the EOF mode with a cumulative variance greater than or equal to 90% can be selected from each EOF mode obtained as the main mode. Thus, the time series corresponding to the main mode, i.e., the principal component, is recorded accordingly.

[0064] S31: Obtaining an eigenvector matrix of the main space matrix according to the time series corresponding to the main space matrix.

[0065] The eigenvector matrix of the main space matrix can be obtained by the following method, including: representing the time series corresponding to the main space matrix as a matrix, calculating the eigenvalues ​​and eigenvectors of the matrix, and obtaining the eigenvector matrix of the main space matrix.

[0066] Specifically, the long time series, i.e., the principal component time series, can be represented as a matrix F, and the covariance matrix of the matrix F is represented as R = F T F, where R is a square matrix. The eigenvalues ​​and eigenvectors of the matrix R can be obtained according to the following formula, namely: RC = CΛ, where C is the eigenvector matrix of the matrix R, which is a diagonal matrix with the elements λ on the diagonal i represents the eigenvalue. The column vector c of the matrix C i is the eigenvector corresponding to the eigenvalue. The covariance matrix R can be decomposed into R = CΛC -1 .

[0067] The eigenvector can be regarded as a set of bases to be found, which represents the characteristics of the original data in the new coordinate system. The eigenvalue in Λ represents the importance of the corresponding feature in the new coordinate system. The eigenvector c1 corresponding to the first eigenvalue, which is the largest eigenvalue, is EOF1, and EOF is obtained in sequence. j . Thus, the eigenvector matrix EOF is obtained.

[0068] If the eigenvector matrix EOF is projected onto the original data, then an EOF represents a quasi-periodic oscillation, and the time variation of this oscillation can be calculated as:

[0069]

[0070] in, Represents the time coefficient or principal component of the eigenvector matrix EOF. By performing power spectrum analysis on the principal component, we can get the time period corresponding to the change of the eigenvector matrix EOF. At the same time, the data can be reconstructed through EOF and principal component, which can be expressed as:

[0071]

[0072] S32: Establish a prediction matrix for the time series corresponding to the main space matrix, multiply the prediction matrix corresponding to the main space matrix by the eigenvector matrix of the main space matrix, and then add them together to construct the third model item.

[0073] The prediction matrix established for the time series corresponding to the main space matrix is ​​used to predict the representation matrix of the time series corresponding to the main space matrix. The representation matrix of the time series refers to the matrix used to represent the time series.

[0074] Optionally, an autoregressive sliding average model can be constructed for the time series corresponding to the main space matrix, so as to construct a prediction matrix for the time series corresponding to the main space matrix. Exemplarily, the prediction matrix of the i-th principal component time series can be expressed as:

[0075] X i (t) = A1X t-1 +…+A p X t-p -B1ε t-1 -…-B q ε t-q .

[0076] Among them, p represents the autoregressive order of the model, q represents the moving average order, A and B represent the non-zero unknown coefficients in the model, and ε represents the independent error term.

[0077] Furthermore, the prediction matrix of the main space matrix time series is multiplied and accumulated with the corresponding eigenvector matrix of the main space matrix, so as to construct a third model term 103, that is, a model term describing the change law of the sea level height residual component of each effective grid in the target area over time.

[0078] Specifically, it can be calculated by matrix multiplication and can be expressed as:

[0079] R(t)=∑R i(t)×EOF i ;

[0080] R i (t) = A1R t-1 +…+A p R t-p -B1ε t-1 -…-B q ε t-q .

[0081] Among them, R(t) represents the third model term, R i (t) represents the prediction matrix of the time series of the i-th main space matrix, EOF i Represents the eigenvector matrix corresponding to the i-th principal space matrix.

[0082] Preferably, in the process of calculating and obtaining the eigenvector matrix of the main space matrix, a significance test can be performed on the obtained modal function, and the error of the kth eigenvalue at a confidence level of 95% can be expressed as: n represents the effective sampling number of data. j ≤λ j+1 -λ j Only after selecting the mode can we decide whether to use the mode as the main mode. This can reduce the number of selected modes.

[0083] In the method of this embodiment, when constructing the third model item 103 for describing the variation pattern of the sea level height residual component of each valid grid in the target area over time, a method combining the empirical orthogonal function (EOF) and the autoregressive moving average (ARMA) model is adopted, which can reduce the computational cost.

[0084] Then, according to the first model item 101, the second model item 102 and the third model item 103 constructed above, the established sea level prediction model can be expressed as:

[0085]

[0086] In practical applications, sea level data of the target area is obtained, and sea level data of the valid grid is obtained therefrom. Preferably, the sea level data of the target area can be input into a base map template for data masking to obtain sea level data of the valid grid. Specifically, the sea level data can be cut according to the base map template of the target area, divided into grids, and the grid representing the land is marked as an invalid grid, and the grid will not be involved in the calculation thereafter.

[0087] This embodiment realizes random dynamic prediction of sea level height by establishing a sea level height prediction model, which has the advantages of a simple model and the ability to reduce calculation costs when predicting sea level height over a large spatial scale.

[0088] In one specific example, Figure 2 This is a schematic diagram of the sea level data at the geographical location (110°E, 5°N) and the first model term established for the trend component. Among them, a univariate linear regression model is established for the sea level data to describe the change pattern of the trend component of the sea level data over time. Figure 3 The figure is a schematic diagram of the sea level data at the geographical location (110°E, 5°N) and the second model term established for the periodic component. Figure 4 The residual part of the sea level data at the geographical location (110°E, 5°N) after removing the trend component and the periodic component is the residual part of the sea level data at the geographical location (110°E, 5°N). It can be seen that the residual part of the sea level data can be regarded as a random sequence.

[0089] Figure 5-1 to Figure 5-6 In order Figure 4 The first 6 principal component time series and predicted values ​​obtained from the data analysis are shown. Figure 6-1 to Figure 6-6 In order Figure 4 The first six main modes obtained from the data analysis are shown.

[0090] Figure 7-1 and Figure 7-2 Comparison between the actual and predicted sea level height in January 2018. Figure 7-1 The sea level height data at the geographical location (110°E, 5°N) in January 2018 measured by a satellite altimeter. Figure 7-2 The sea level data at the geographical location (110°E, 5°N) in January 2018 predicted by the method of this embodiment is shown in Figure 1. It can be seen that although there are some deviations in the predicted values ​​of some sea areas, with an average absolute error of 50.5 mm, the predicted results are similar to the true values ​​in terms of spatial distribution trend, so the model can be used for sea level prediction.

[0091] Correspondingly, this embodiment also provides a sea level prediction system for executing the above-mentioned sea level prediction method.

[0092] The sea level prediction system of this embodiment realizes random dynamic prediction of sea level by establishing a sea level prediction model. Compared with the prior art, it can reduce the amount of calculation in predicting sea level on a large spatial scale and improve the prediction accuracy.

[0093] Optionally, the sea level prediction system of this embodiment includes: an acquisition device for acquiring the sea level prediction model for the target area; and a prediction device for using the sea level prediction model to obtain sea level data of each valid grid in the target area at a preset time.

[0094] Correspondingly, this embodiment further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the sea level prediction method as described above are implemented.

[0095] The above is a detailed introduction to a sea level prediction method and system, and a readable storage medium provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention, and the description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting sea level, characterized in that: Using a sea level prediction model established for a target area, obtaining sea level data of each valid grid in the target area at a preset time; The sea level prediction model includes a first model item, a second model item and a third model item, wherein the first model item describes the variation law of the sea level trend component of each effective grid in the target area over time, the second model item describes the variation law of the sea level periodic component of each effective grid in the target area over time, and the third model item describes the variation law of the sea level residual component of each effective grid in the target area over time, the sea level trend component refers to the part of the sea level that shows a linear variation trend over time, the sea level periodic component refers to the part of the sea level that shows a periodic variation over time, and the sea level residual component refers to the part of the sea level other than the trend component and the periodic component; Constructing the third model term includes: Obtain the main space matrix of the portion of the sea level height of each effective grid in the target area minus the trend component and the periodic component and the corresponding time series; According to the time series corresponding to the main space matrix, the characteristic vector matrix of the main space matrix is ​​obtained, including: the time series corresponding to the main space matrix is ​​represented as a matrix F, and the covariance matrix of the matrix F is represented as R=F T F, the eigenvalues ​​and eigenvectors of the matrix R are obtained according to the following formula, namely: RC = CΛ, where C is the eigenvector matrix of the matrix R, and the eigenvector c1 corresponding to the largest eigenvalue in Λ is the eigenvector matrix EOF1 of the main space matrix; A prediction matrix is ​​established for the time series corresponding to the main space matrix, wherein the prediction matrix is ​​used to predict the representation matrix of the time series corresponding to the main space matrix, and the representation matrix of the time series refers to a matrix used to represent the time series; The prediction matrix corresponding to the main space matrix is ​​multiplied and added with the eigenvector matrix of the main space matrix to construct the third model term, which is expressed as: R(t)=∑R i (t)×EOF i ;R i (t)=A1R t-1 +…+A p R t-p -B1ε t-1 -…-B q ε t-q ; Among them, R(t) represents the third model term, R i (t) represents the prediction matrix of the time series of the i-th main space matrix, p represents the autoregressive order of the model, q represents the moving average order, A and B represent the non-zero coefficients in the model, ε represents the independent error term, EOF i Represents the eigenvector matrix corresponding to the i-th principal space matrix.

2. The sea level prediction method according to claim 1, characterized in that: The first model term describes the change of the sea level height trend component of each effective grid in the target area over time by a linear function.

3. The sea level prediction method according to claim 1, characterized in that: The second model term describes the change of the periodic component of the sea level height of each effective grid in the target area over time by using a sine function and / or a cosine function.

4. The sea level prediction method according to claim 3, characterized in that: Constructing the second model item includes: performing spectral analysis on the part obtained by subtracting the trend component from the sea level height of each valid grid in the target area, selecting a main period from the obtained periods according to the amplitude values ​​corresponding to the obtained periods, and constructing the second model item with the sine function and / or cosine function represented by the main period.

5. The method for predicting sea level according to claim 1, characterized in that: The third model term includes a model obtained by multiplying and accumulating a prediction matrix of the main space matrix time series of the sea level height residual component of each valid grid in the target area and a eigenvector matrix of the main space matrix.

6. A sea level prediction system, characterized in that: Used to execute the sea level prediction method described in any one of claims 1-5.

7. The sea level prediction system according to claim 6, characterized in that: include: An acquisition device, used to acquire the sea level height prediction model for the target area; The prediction device is used to use the sea level prediction model to obtain the sea level data of each valid grid in the target area at a preset time.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the sea level prediction method according to any one of claims 1 to 5 are implemented.

Images