Area estimation method for marine environmental dynamics

Abstract

The present invention belongs to the field of coastal and marine engineering technology, and discloses a method for estimating the area of ​​marine environmental dynamic elements, which mainly solves the problem of large uncertainty in design basis estimates of marine and coastal structures in different return periods due to the lack of historical data of marine environmental dynamic element samples in some sea area positions. [Solution] The area estimation method for marine environment dynamic elements of the present invention includes the steps of acquiring n-year time series historical data of marine environment dynamic elements, grouping by meteorological system and extracting independent and identically distributed extreme samples, calculating each probability distribution parameter for single points in areas with sufficient information on extreme samples, creating a spatial development model of each parameter and longitude, latitude and water depth based on an artificial neural network, correcting the probability distribution parameters of single points in areas with sufficient extreme samples based on the created model, extrapolating the spatial model to areas with insufficient information on extreme samples, and calculating confidence intervals of area model estimates for different return periods using an area re-simulation method.

Description

[Technical field]

[0001] The present invention relates to the field of coastal and marine engineering technology, and more particularly to a method for estimating the area of ​​marine environmental dynamics elements. [Background technology]

[0002] In recent years, with the gradual optimization of traditional marine industries and the continuous innovation of high-tech marine industries, the development of marine and coastal engineering has faced new opportunities as well as new challenges. Marine environmental dynamics factors (such as extreme significant wave height) are important parameters in coastal structural design and management, offshore platforms, shipping lines and port construction.

[0003] Typhoons are serious natural disaster phenomena that pose a great threat to marine and coastal construction. In areas where typhoons occur relatively frequently, sufficient historical data is available, and it is certain to obtain design values ​​of marine environmental dynamics factors using traditional estimation methods. In areas where typhoons occur infrequently (annual frequency less than 1), the marine environmental dynamics factors associated with them are extreme, so their impact cannot be ignored when designing marine and coastal structures. However, due to the limited historical data, there is a large uncertainty in obtaining design values ​​of marine environmental dynamics factors using traditional estimation methods, and they cannot meet the needs of actual construction.

[0004] In recent years, the "sky, earth, sea" ocean observation system has been gradually formed, ocean observation and simulation data has shown exponential growth, and ocean data has already entered the era of big data. However, in some sea areas, the historical data is still short, and the design values ​​of marine environmental dynamic elements obtained by traditional estimation methods have large uncertainties, which seriously restrict the construction and development of marine engineering. Summary of the Invention [Problem to be solved by the invention]

[0005] There exists a problem of large uncertainties in design basis estimates for marine and coastal structures at different return periods due to a lack of historical data on marine environmental dynamics samples in some sea areas. [Means for solving the problem]

[0006] A method for estimating the area of ​​a marine environmental dynamics element for solving the above problem includes the steps of: Step S1 of acquiring n-year (n is greater than 20 years, and the marine dynamic environmental element is significant wave height) time series history data of the marine dynamic environmental element; A step S2 of extracting independent and identically distributed extreme samples grouped by weather system; In an area where information on extreme samples is sufficient, a step S3 of calculating each probability distribution parameter for a single point and applying the GPD distribution model to each ocean area point to calculate each probability distribution parameter; Step S4 of creating a spatial evolution model of each parameter and the longitude, latitude and water depth based on an artificial neural network; Step S5: based on the created model, correct the probability distribution parameters of single points in the extreme sample sufficiency area, i.e., apply the model to each position in the extreme sample sufficiency area to obtain three parameters of the GPD distribution as corrected values ​​instead of the original values; Step S6: extrapolating the spatial model to a location in an area where the information of extreme samples is insufficient, i.e., an extremely sample-deficient area, and gradually extrapolating to a location in the extremely sample-deficient area using a parameter evolution mode of the spatial model; and a step S7 of calculating confidence intervals for the estimates of the area model at different return periods by the area resimulation method, In step S4, Step S41 of constructing a neural network model including a hidden layer using a hyperbolic tangent sigmoid type transfer function, an input layer using a linear transfer function, and an output layer, and performing data training using the Levenberg-Marquardt backpropagation method; Step S42 of setting the output layer of the model as parameters of a probability distribution model; TIFF2025513666000002.tif11170

[0007] Furthermore, step S2 specifically includes the following steps: Step S21: for a meteorological system with prominent typhoon characteristics and complete information records, matching is performed between trajectory tracking and recording time based on trajectory, impact range and recording time information, and independent and identically distributed samples are extracted; and a step S22 of extracting independent samples at regular time intervals for other meteorological systems and extracting identically distributed samples using directional grouping and time grouping means.

[0008] Furthermore, step S4 specifically includes the following steps: Step S41 of constructing a neural network model including a hidden layer using a hyperbolic tangent sigmoid type transfer function, an input layer using a linear transfer function, and an output layer, and performing data training using the Levenberg-Marquardt backpropagation method; Step S42: setting the output layer of the model to the parameters of the probability distribution model, and determining the input layer parameters of the neural network model including longitude, latitude, and water depth based on three statistical indices: root mean square error (RMSE), mean absolute error (MAE), and correlation coefficient (R); TIFF2025513666000003.tif11170

[0009] TIFF2025513666000004.tif214170 Effect of the Invention

[0010] The present invention takes the significant wave height under the influence of tropical cyclones as an example, and uses the peak super-threshold method of flexible time window to extract extreme samples, ensuring that the samples meet the independent and identical distribution requirements of extreme value theory. An area extrapolation model based on artificial neural network is constructed by analyzing the three-dimensional evolution relationship between the probability distribution parameters of each single point in the area and the longitude, latitude and water depth, so as to control the local response of the probability distribution parameters, and a spatial control law is constructed based on the characteristics of the single point and the wave characteristics, so as to increase the effective sample information for the estimation of the return period of the single point, and exchange time with space, thereby reducing the uncertainty of the estimation of marine environmental dynamic elements and extending the estimation period. This area sampling method can provide important reference for the design and construction of marine and coastal engineering.

[0011] This method mainly solves the problem of large uncertainty of design basis estimates of marine and coastal structures in different return periods caused by the lack of historical data of marine environmental dynamic element samples in some sea area locations. This method establishes spatial control laws of probability distribution parameters based on location characteristics and wave characteristics to control the local response of probability distribution parameters, and increases the effective sample information for estimating return periods. Compared with traditional estimation methods, this method can obtain more reliable estimates of design basis of marine environmental dynamic elements in different return periods. This innovative measure of exchanging time in space can effectively extend the estimated period, provide important reference for the design and construction of marine and coastal engineering works, and meet the actual construction needs. [Brief description of the drawings]

[0012] In order to more clearly describe the embodiments of the present invention or the technical solutions in the prior art, the drawings that need to be used in the description of the embodiments or the prior art are briefly described below. Obviously, the drawings in the following description are only the embodiments of the present invention, and those skilled in the art can obtain other drawings based on the drawings provided without paying creative labor. [Figure 1] FIG. 1 is a flow diagram of the method for estimating the area of ​​a marine environment driving element disclosed in the present invention. [Diagram 2] FIG. 2 shows the overall average difference (RMSE and MAE statistics) between the distribution curve in the area in the extreme sample sufficiency area and the integrated distribution curve obtained by integrating the area model at a single point in the embodiment. [Diagram 3] FIG. 3 is a time-series diagnostic diagram of the fitted spatial model GPD parameters of extreme samples at the A2 position in the embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] In an embodiment of the present invention, a method for estimating the area of ​​marine environmental dynamics elements is disclosed, in which the reference points are A1 (124.0°E, 28.0°N), A2 (123.6°E, 36.3°N), and A3 (123.6°E, 38.8°N), among which point A1 is located in the northern sea area of ​​the East Sea, point A2 is located in the central sea area of ​​the Yellow Sea, and point A3 is located in the northern sea area of ​​the Yellow Sea. Point A1 belongs to a location with sufficient extreme sample information, point A2 is located in the southern part of the Yellow Sea, which is lacking in extreme typhoon samples, and point A3 is located in the northern part of the Yellow Sea, which is even lacking in 40-year extreme typhoon samples.

[0014] In this embodiment, the following steps are included. In step S1, n-year time series history data of marine environmental dynamic elements is obtained. The marine dynamic environmental element is significant wave height, n is 40 years, the time range is 1979-2018, and the time step is 1h.

[0015] In step S2, weather systems are grouped and independent and identically distributed extreme samples are extracted.

[0016] In step S21, for a weather system with prominent typhoon characteristics and complete information records, matching is performed between trajectory tracking and recording time based on trajectory, impact range, and recording time information, and independent and identically distributed samples are extracted. In this embodiment, since the significant wave height under the influence of a typhoon is analyzed, matching is performed between trajectory tracking and recording time based on information such as the trajectory impact range (300 km) and recording time of each typhoon within 40 years, and independent and identically distributed samples are extracted. As a result of screening, in this embodiment, 178 samples of point A1, 74 samples of point A2, and 47 samples of point A3 were extracted in 40 years.

[0017] In step S22, independent samples are extracted at regular time intervals for other meteorological systems, and samples with identical distribution are extracted using means such as directional grouping and time grouping.

[0018] In step S3, in an area with sufficient information on extreme samples, each probability distribution parameter is calculated for each location. In this embodiment, as a result of the analysis, point A1 belongs to a location with sufficient information on extreme samples, and the number of extreme samples is greater than once a year. Therefore, in an area with sufficient information on extreme samples, the GPD distribution model is applied to each sea area point to calculate each probability distribution parameter. Sea area points A2 and A3 belong to locations with insufficient extreme samples.

[0019] In step S4, a spatial evolution model of each parameter and the longitude, latitude and water depth is created based on an artificial neural network.

[0020] TIFF2025513666000005.tif36170

[0021] TIFF2025513666000006.tif62170 [Table 1] [Table 2]

[0022] TIFF2025513666000009.tif49170

[0023] In step S5, the probability distribution parameters of the single points in the extreme sample sufficiency area are corrected based on the created model. In this embodiment, based on the created area model, the model is applied to each position in the extreme sample sufficiency area to obtain three parameters of the GPD distribution, which are used as the corrected values ​​instead of the original values.

[0024] In step S6, the spatial model is extrapolated to areas with insufficient information on extreme samples. In this embodiment, in the area positions (A2 and A3) where extreme samples are insufficient, the historical data of extreme samples is insufficient, so reasonable GPD distribution parameters cannot be obtained by estimating their own historical data, so the parameter development mode using the spatial model is gradually extrapolated to the area positions where extreme samples are insufficient. Figure 3 shows a comparison diagram between the quantiles calculated from the three parameters of the GPD distribution obtained by the spatial model at position A2 and the empirical quantiles of the original samples, which can be seen to be relatively consistent, indicating the reliability of the spatial model.

[0025] In step S7, confidence intervals for the estimates of the area model at different return periods are calculated by the area resimulation method.

[0026] TIFF2025513666000010.tif25170

[0027] TIFF2025513666000011.tif98170

[0028] TIFF2025513666000012.tif84170

[0029] TIFF2025513666000013.tif36170

[0030] According to an embodiment of the present invention, at positions A2 and A3, the historical data of extreme typhoons is too few compared to the historical period (40 years), so there is a large uncertainty when estimating the design wave height of marine and coastal structure construction, and there is also a problem that the typhoon samples are too few inside the Bohai Sea, which affects the future defense against typhoons of marine construction. However, by applying this area estimation method, the problem of large uncertainty in estimating the design wave height and even the inability to estimate it can be effectively solved, providing important reference for the design and construction of marine and coastal construction, and meeting the defense requirements against extreme disasters such as typhoons in actual construction. Table 3 shows the comparative results of estimating the stability of design wave height by applying the traditional single point estimation method and the area estimation method at locations A1, A2, and A3, and by the traditional single point estimation method, the confidence intervals of the once in 100-year design wave height are 1.706 m, 2.381 m, and 3.534 m, respectively, while by the area estimation method, the confidence intervals of the once in 100-year design wave height are 1.684 m, 2.089 m, and 1.988 m, respectively. Point A1 is located in the north of the East Sea, where the samples of extreme typhoons in 40 years are sufficient, and the area estimation method reduces the confidence interval by 0.022m compared to the traditional single-point estimation; point A2 is located in the south of the Yellow Sea, where the samples of extreme typhoons are insufficient, and the area estimation method reduces the confidence interval by 0.292m compared to the traditional single-point estimation; point A3 is located in the north of the Yellow Sea, where the samples of extreme typhoons in 40 years are even less, and the area estimation method reduces the confidence interval by 1.546m compared to the traditional single-point estimation, which is nearly one-fold. Therefore, by applying this method, a more reliable design wave height estimate can be obtained, which can provide important reference for the design and construction of marine and coastal engineering works, and meet the needs of typhoon disaster prevention in actual engineering works. [Table 3]

[0031] Each embodiment in this specification is described step by step, and each embodiment is described with emphasis on the difference from other embodiments, and the same or similar parts between each embodiment may be referred to each other. The apparatus disclosed in the embodiment corresponds to the method disclosed in the embodiment, so it is relatively simple to describe, and the related points may be referred to the description of the method part.

[0032] The above description of the disclosed embodiments enables those skilled in the art to realize or use the present invention. Various modifications of these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to these embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. Step S1 of acquiring n-year (n is greater than 20 years, and the marine dynamic environmental element is significant wave height) time series history data of the marine dynamic environmental element; A step S2 of extracting independent and identically distributed extreme samples grouped by weather system; In an area where information on extreme samples is sufficient, a step S3 of calculating each probability distribution parameter for a single point and applying a GPD distribution model to each sea area point to calculate each probability distribution parameter; Step S4 of creating a spatial evolution model of each parameter and the longitude, latitude and water depth based on an artificial neural network; Step S5: based on the created model, correct the probability distribution parameters of single points in the extreme sample sufficiency area, i.e., apply the model to each position in the extreme sample sufficiency area to obtain three parameters of GPD distribution as corrected values ​​instead of the original values; Step S6: extrapolating the spatial model to a location in an area where the information of extreme samples is insufficient, i.e., an extremely sample-deficient area, and gradually extrapolating to a location in the extremely sample-deficient area using a parameter evolution mode of the spatial model; and a step S7 of calculating confidence intervals of the area model estimates for different return periods by the area resimulation method, The step S4 is Step S41 of constructing a neural network model including a hidden layer using a hyperbolic tangent Sigmoid type transfer function, an input layer using a linear transfer function, and an output layer, and performing data training using the Levenberg-Marquardt backpropagation method; Step S42: setting the output layer of the model as a parameter of the probability distribution model; A method for estimating the area of ​​marine environmental dynamics elements.

2. Specifically, step S2 is Step S21: for a meteorological system with prominent typhoon characteristics and complete information records, matching is performed between trajectory tracking and recording time based on trajectory, impact range and recording time information, and independent and identically distributed samples are extracted; and (S22) extracting independent samples at regular time intervals for other meteorological systems and extracting identically distributed samples using directional grouping and time grouping means. The method for estimating the area of ​​a marine environment dynamics element according to claim 1 .

3. In step S42, the output layer of the model is set to the parameters of the probability distribution model, and the input layer parameters of the neural network model including the longitude, latitude, and water depth are determined according to three statistical indices, namely, the root mean square error RMSE, the mean absolute error MAE, and the correlation coefficient R; The method for estimating the area of ​​a marine environment dynamics element according to claim 2 .

4.

Images