Polar region sea ice concentration prediction method and device

Abstract

The invention discloses a polar region sea ice concentration prediction method and device, and belongs to the technical field of climatic sciences, and the method comprises the steps: obtaining the frequency spectrum information of the sea ice concentration through a historical year-round daily polar region high temporal-spatial resolution image set, building a sea ice concentration prediction model based on the frequency spectrum information of the sea ice concentration, and carrying out the prediction of the sea ice concentration. The prediction model is constructed only based on the annual high temporal-spatial resolution image set, sample data is reliable and does not involve long-time-dimension data calculation amount, the calculation complexity of the model is reduced, the prediction precision is improved, and the problems that a current sea ice prediction model is high in calculation complexity and short in prediction period are solved.

Description

Technical Field

[0001] The present invention relates to a method and device for predicting polar sea ice concentration, belonging to the technical field of climate science and technology. Background Art

[0002] Sea Ice Concentration (SIC) is one of the core indicators in the study of polar climate change. The growth and ablation of sea ice have a profound impact on the global climate system, including affecting multiple fields such as ocean circulation, climate modeling, ecosystems, and human activities; the change of sea ice concentration is particularly significant in polar regions and is directly closely related to polar climate change, global sea level change, and species habitat change, etc. Therefore, accurately predicting the concentration of polar sea ice is of extremely important significance.

[0003] Most of the existing sea ice prediction methods rely on physics-based models (such as numerical climate models) or deep learning models (such as convolutional neural networks). Although these methods have made progress to a certain extent, physical models usually involve a large number of complex numerical simulations, consume a large amount of computing resources, and require high-performance computers for long-time calculations, and the computational complexity of the models is high; deep learning models have a short prediction period and rely on continuous sea ice prediction data and cannot provide long-term sea ice change trends. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art, provide a method and device for predicting polar sea ice concentration, reduce the computational complexity of the model and improve the prediction accuracy, and solve the problems of high computational complexity and short prediction period of the current sea ice prediction model.

[0005] To solve the above technical problems, the present invention is implemented by adopting the following technical solutions: A method for predicting polar sea ice concentration, comprising: Obtaining a target time node; Inputting the target time node into a pre-constructed sea ice concentration prediction model to obtain a predicted value of polar sea ice concentration corresponding to the time node; Obtaining the pre-constructed sea ice concentration prediction model, comprising: Obtaining the spectral information of sea ice concentration according to the historical annual daily polar high spatio-temporal resolution atlas; Determining the frequency domain signal of the original sequence of the corresponding sea ice concentration according to the spectral information of sea ice concentration; Based on the frequency domain signal, establishing an initial sea ice concentration prediction model; Fitting the image corresponding to the initial sea ice concentration prediction model based on the historical annual daily polar high spatio-temporal resolution atlas, and determining the optimal model parameters that match the fitted image; Replace the model parameters of the initial sea ice concentration prediction model with the optimal model parameters to obtain the constructed sea ice concentration prediction model.

[0006] Furthermore, the obtaining of the spectral information of the sea ice concentration according to the historical high spatio-temporal resolution atlas of the polar region for each day of the whole year includes: Obtain the average value of the sea ice concentration at the first spatial resolution for each day of the whole year in the polar region according to the historical high spatio-temporal resolution atlas of the polar region for each day of the whole year; Perform discrete Fourier transform processing on the average value of the sea ice concentration at the first spatial resolution for each day of the whole year in the polar region to obtain the spectral information corresponding to the sea ice concentration.

[0007] Even further, the single picture coverage in the historical high spatio-temporal resolution atlas of the polar region for each day of the whole year includes the polar region scope.

[0008] Even further, the first spatial resolution includes the spatial resolution.

[0009] Furthermore, the establishing of the initial sea ice concentration prediction model based on the frequency domain signal includes: Obtain the amplitude and phase of each frequency component of the frequency domain signal, and screen out the frequency components that meet the modeling requirements according to the amplitude and phase of each frequency component of the frequency domain signal; Determine the initial sea ice concentration prediction model according to the frequency components that meet the modeling requirements.

[0010] Even further, the determining of the initial sea ice concentration prediction model according to the frequency components that meet the modeling requirements includes: The initial sea ice concentration prediction model is expressed as: ; where: represents the predicted value of the sea ice concentration, represents the time node; A represents the first-order term model parameter, B represents the constant term model parameter, C represents the first sine term model parameter, D represents the second sine term model parameter; represents the first periodic feature, represents the second periodic feature, corresponds to , represents the first frequency component, , represents the second frequency component, ; represents pi; represents the sine function.

[0011] Furthermore, for the image corresponding to the initial sea ice concentration prediction model fitted based on the historical annual daily high spatio-temporal resolution atlas of the polar region, determining the optimal model parameters matching the fitted image includes: Obtaining the change trend of each pixel point representing the sea ice concentration in a single picture of the historical annual daily high spatio-temporal resolution atlas of the polar region throughout the year; Based on the change trend, determining the optimal model parameters of the initial sea ice concentration prediction model.

[0012] Furthermore, for determining the optimal model parameters of the initial sea ice concentration prediction model based on the change trend, it includes: Using the least squares method to perform pixel-by-pixel fitting on the single picture in the atlas, and determining the model parameters of the initial sea ice concentration prediction model corresponding to each pixel point; Through the initial sea ice concentration prediction model with the determined model parameters, obtaining the predicted sea ice concentration values of each pixel point at historical time nodes; Based on the predicted sea ice concentration values and the actual values at the same historical time nodes, determining the residual sequence representing the change trend; Based on the residual sequence, correcting the model parameters of the initial sea ice concentration prediction model to obtain the optimal model parameters of the initial sea ice concentration prediction model.

[0013] Furthermore, for determining the residual sequence representing the change trend based on the predicted sea ice concentration values and the actual values at the same historical time nodes, it includes: Calculating the residual value using the following formula: ; where: represents the residual value, represents the predicted value output by the initial sea ice concentration prediction model at the i th time node, represents the actual value of the sea ice concentration at the i th time node; Sorting the time nodes corresponding to each residual value to form a residual sequence.

[0014] The present invention also provides a polar sea ice concentration prediction device, including: An acquisition module, configured to acquire the target time node; A processing module, configured to input a target time node into a pre-constructed sea ice concentration prediction model to obtain a predicted value of polar sea ice concentration corresponding to the time node; Obtaining the pre-constructed sea ice concentration prediction model includes: According to the historical annual daily polar high spatio-temporal resolution atlas, obtaining the spectral information of sea ice concentration; According to the spectral information of sea ice concentration, determining the frequency domain signal of the original sequence of the corresponding sea ice concentration; Based on the frequency domain signal, establishing an initial sea ice concentration prediction model; Based on the historical annual daily polar high spatio-temporal resolution atlas, fitting the image corresponding to the initial sea ice concentration prediction model, and determining the optimal model parameters that match the fitted image; Replacing the model parameters of the initial sea ice concentration prediction model with the optimal model parameters to obtain the constructed sea ice concentration prediction model.

[0015] Compared with the prior art, the beneficial effects achieved by the present invention are: 1. The present invention obtains the spectral information of sea ice concentration through the historical annual daily polar high spatio-temporal resolution atlas. Based on the spectral information of sea ice concentration, a sea ice concentration prediction model is constructed. The prediction model is constructed only based on the annual high spatio-temporal resolution atlas. The sample data is reliable and does not involve the computational amount of data in the long time dimension, reducing the computational complexity of the model and improving the prediction accuracy, and solving the problems of high computational complexity and short prediction period of the current sea ice prediction model.

[0016] 2. The present invention obtains the change trend of each pixel point representing sea ice concentration in a single picture of the historical annual daily polar high spatio-temporal resolution atlas during a year, and based on the change trend, determines the optimal model parameters of the initial sea ice concentration prediction model, ensuring the reliability of the relevant parameters of the prediction model and greatly improving the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a flowchart of a method for predicting polar sea ice concentration provided by an embodiment of the present invention; Figure 2 is a flowchart of constructing a sea ice concentration prediction model provided by an embodiment of the present invention; Figure 3 is a schematic diagram of the annual daily sea ice concentration mean sequence provided by an embodiment of the present invention (the abscissa in the figure represents the number of days, and the ordinate represents the daily sea ice concentration mean); Figure 4 is a schematic diagram of the spectral information corresponding to the annual daily sea ice concentration mean sequence provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and cannot be used to limit the protection scope of the present invention. Embodiment

[0019] As Figure 1 shown, a method for predicting polar sea ice concentration includes: Obtain the spectral information of sea ice concentration according to the historical annual daily polar high spatio-temporal resolution atlas. Specifically: As Figure 2 shown, obtain the average value of sea ice concentration at the first spatial resolution of the polar region for each day of the year according to the historical annual daily polar high spatio-temporal resolution atlas. In this embodiment, the data of the historical annual daily polar high spatio-temporal resolution atlas provided by the University of Bremen is used. The atlas data covers the period from 2013 to 2024, and the single picture of the atlas covers the polar region including The first spatial resolution includes The spatial resolution of Perform discrete Fourier transform processing on the average value of sea ice concentration at the first spatial resolution of the polar region for each day of the year to obtain the spectral information corresponding to the sea ice concentration.

[0020] Based on the spectral information of sea ice concentration, construct a sea ice concentration prediction model. Specifically: Determine the frequency domain signal of the original sequence of the corresponding sea ice concentration according to the spectral information of the sea ice concentration; Obtain the amplitude and phase of each frequency component of the frequency domain signal, and screen out the frequency components that meet the modeling requirements according to the amplitude and phase of each frequency component of the frequency domain signal; It should be noted that the daily average sea ice concentration data is converted from a time domain signal to a frequency domain signal. The frequency signal can show the amplitude and phase distribution of the signal at different frequencies. Since the output result of the Fourier transform is a complex number, its modulus represents the amplitude of the frequency component, and the magnitude of the amplitude determines the contribution of the frequency component in the original sequence; As Figure 3 And Figure 4 shown, the amplitude range of the original sea ice concentration sequence is 0 to 30, the maximum frequency amplitude is 5, the frequency with the largest contribution rate in the original amplitude is 0.0027 hz, followed by 0.0055 hz. For the integrity of the model, two frequency components with larger contribution rates are selected for modeling, namely the first frequency component , the second frequency component , and the initial sea ice concentration prediction model is expressed as: ; Where: Represents the predicted value of sea ice concentration, Represents a time node; A Represents the first-order term model parameter, B Represents the constant term model parameter, C Represents the first sine term model parameter, D Represents the second sine term model parameter; Represents the first periodic feature, Represents the second periodic feature, Corresponds to , Represents the first frequency component, , Corresponds to 365.2 days, Represents the second frequency component, , Corresponds to 182.6 days; Represents pi; Represents the sine function.

[0021] Obtain the change trend of each pixel point representing sea ice concentration in a single picture of the historical annual daily high spatiotemporal resolution atlas of the polar region throughout the year; Based on the said change trend, determine the optimal model parameters of the initial sea ice concentration prediction model, including: Adopt the least squares method to fit each pixel in the single picture of the atlas, and determine the model parameters of the initial sea ice concentration prediction model corresponding to each pixel point; Through the initial sea ice concentration prediction model with the determined model parameters, obtain the predicted value of sea ice concentration at each pixel point at the historical time node; Based on the predicted value and the actual value of sea ice concentration at the same historical time node, determine the residual sequence representing the change trend, and calculate the residual value using the following formula: ; Where: Represents the residual value, Represents the i th predicted value output by the initial sea ice concentration prediction model at the time node, Represents the i th actual value of sea ice concentration at the time node; Sort the time nodes corresponding to each residual value to form a residual sequence and correct the model parameters of the initial sea ice concentration prediction model according to the residual sequence. In this embodiment, the optimal model parameter values are: , , , ; Replace the model parameters of the initial sea ice concentration prediction model with the optimal model parameters to obtain the constructed sea ice concentration prediction model, which is expressed as: ; Input the target time node into the constructed sea ice concentration prediction model to obtain the predicted value of the polar sea ice concentration at the corresponding time node.

[0022] To verify the prediction effect of the method proposed in this embodiment, the mean square error is used as the verification index, and the calculation formula of the mean square error is: ; where, represents the mean square error value, n represents the number of time nodes; The calculation results of the mean square error value are shown in the following table: Table 1 As can be seen from Table 1, the prediction error of the method proposed by the present invention generally shows a downward trend, with the minimum error being 0.0659 around 200 days, and there is a slight increase around 15 days and 250 days due to the influence of seasons. Therefore, the method proposed in this embodiment has high prediction accuracy. Embodiment

[0023] The present invention also provides a polar sea ice concentration prediction device, including: An acquisition module for acquiring the target time node; A processing module for inputting the target time node into the constructed sea ice concentration prediction model to obtain the predicted value of the polar sea ice concentration at the corresponding time node; obtaining the constructed sea ice concentration prediction model, including: According to the historical annual daily polar high spatio-temporal resolution atlas, obtain the spectral information of the sea ice concentration; According to the spectral information of the sea ice concentration, determine the frequency domain signal of the corresponding original sea ice concentration sequence; Based on the frequency domain signal, establish an initial sea ice concentration prediction model; Based on the historical annual daily polar high spatio-temporal resolution atlas, fit the image corresponding to the initial sea ice concentration prediction model, and determine the optimal model parameters that match the fitted image; Replace the model parameters of the initial sea ice concentration prediction model with the optimal model parameters to obtain the constructed sea ice concentration prediction model.

[0024] In the description of the present invention, it should be understood that terms such as "first", "second", etc. are only used for descriptive purposes and cannot be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", etc. may explicitly or implicitly include one or more of such features.

[0025] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0026] The present application is described with reference to the flowcharts of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process in the flowchart can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the functions specified in Figure 1 one process or multiple processes or boxes Figure 1 one box or multiple boxes.

[0027] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction device that implements the functions specified in Figure 1 one process or multiple processes.

[0028] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, thereby providing steps for implementing the functions specified in Figure 1 one process or multiple processes.

[0029] The embodiments of the present invention have been described above in conjunction with the accompanying drawings. However, the present invention is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not restrictive. Under the inspiration of the present invention, those of ordinary skill in the art can also make many forms without departing from the spirit and scope protected by the present invention and the claims. These all fall within the protection scope of the present invention.

Claims

1. A method for predicting polar sea ice density, characterized in that: include: Get the target time node; Input the target time node into the constructed sea ice density prediction model to obtain the polar sea ice density prediction value at the corresponding time node; Obtaining the constructed sea ice density prediction model, including: Obtain spectral information on sea ice density based on the historical full-year daily polar high temporal and spatial resolution atlas; According to the spectrum information of the sea ice density, a frequency domain signal corresponding to the original sequence of the sea ice density is determined; Based on the frequency domain signal, an initial sea ice density prediction model is established; Fit the image corresponding to the initial sea ice density prediction model based on the historical annual daily polar high temporal and spatial resolution atlas, and determine the optimal model parameters that match the fitted image; The model parameters of the initial sea ice density prediction model are replaced by the optimal model parameters to obtain the constructed sea ice density prediction model.

2. The method for predicting polar sea ice density according to claim 1, characterized in that: The spectrum information of sea ice density obtained based on the historical annual daily polar high temporal and spatial resolution atlas includes: Based on the historical polar high temporal and spatial resolution atlas, the mean polar sea ice density at the first spatial resolution is obtained. The mean value of the polar sea ice density at the first spatial resolution every day throughout the year is processed by discrete Fourier transform to obtain the spectral information corresponding to the sea ice density.

3. The polar sea ice density prediction method according to claim 2, characterized in that: The single image coverage of the historical annual daily polar high temporal and spatial resolution atlas includes polar region.

4. The method for predicting polar sea ice density according to claim 2, characterized in that: The first spatial resolution includes spatial resolution.

5. The method for predicting polar sea ice density according to claim 1, characterized in that: The initial sea ice density prediction model is established based on the frequency domain signal, including: Obtain the amplitude and phase of each frequency component of the frequency domain signal, and filter out the frequency components that meet the modeling requirements according to the amplitude and phase of each frequency component of the frequency domain signal; The initial sea ice concentration prediction model is determined based on the frequency components that meet the modeling requirements.

6. The method for predicting polar sea ice density according to claim 5, characterized in that: The initial sea ice density prediction model is determined according to the frequency components that meet the modeling requirements, including: The initial sea ice density prediction model is expressed as: ； in: represents the predicted value of sea ice concentration, Indicates a time node; A represents the linear model parameter, B represents the constant term model parameter, C represents the first sinusoidal model parameter, D represents the second sinusoidal model parameter; represents the first period characteristics, represents the second periodic characteristic, correspond , represents the first frequency component, , represents the second frequency component, ; represents pi; Represents the sine function.

7. The method for predicting polar sea ice density according to claim 5, characterized in that: The image corresponding to the initial sea ice density prediction model is fitted based on the historical annual daily polar high temporal and spatial resolution atlas, and the optimal model parameters matching the fitted image are determined, including: Obtain the changing trend of each pixel representing the sea ice density in a single image of the historical daily polar high temporal and spatial resolution atlas throughout the year; Based on the changing trend, optimal model parameters of the initial sea ice concentration prediction model are determined.

8. The method for predicting polar sea ice density according to claim 7, characterized in that: Determining optimal model parameters of the initial sea ice density prediction model based on the change trend includes: The least square method is used to fit the single image in the atlas pixel by pixel to determine the model parameters of the initial sea ice density prediction model corresponding to each pixel point; The predicted sea ice density value of each pixel at a historical time node is obtained through an initial sea ice density prediction model with determined model parameters; Based on the predicted and actual values ​​of sea ice density at the same historical time node, the residual sequence representing the change trend is determined; The model parameters of the initial sea ice concentration prediction model are corrected based on the residual sequence, and the optimal model parameters of the initial sea ice concentration prediction model are obtained.

9. The method for predicting polar sea ice density according to claim 8, characterized in that: The residual sequence representing the change trend is determined based on the predicted value and the actual value of sea ice density at the same historical time node, including: The residual value is calculated using the following formula: ； in: represents the residual value, Indicates i The predicted value output by the initial sea ice density prediction model at each time node, Indicates i The actual value of sea ice density at each time point; Sort each residual value corresponding to the time node to form a residual sequence.

10. A polar sea ice density prediction device, characterized in that: include: An acquisition module is used to acquire a target time node; A processing module is used to input the target time node into the constructed sea ice density prediction model to obtain the polar sea ice density prediction value of the corresponding time node; Obtaining the constructed sea ice density prediction model, including: Obtain spectral information on sea ice density based on the historical full-year daily polar high temporal and spatial resolution atlas; According to the spectrum information of the sea ice density, a frequency domain signal corresponding to the original sequence of the sea ice density is determined; Based on the frequency domain signal, an initial sea ice density prediction model is established; Fit the image corresponding to the initial sea ice density prediction model based on the historical annual daily polar high temporal and spatial resolution atlas, and determine the optimal model parameters that match the fitted image; The model parameters of the initial sea ice density prediction model are replaced by the optimal model parameters to obtain the constructed sea ice density prediction model.

Images