Polar sea ice density inversion method and device
The GNSS-R DDM data and sea ice density field data are processed through the convolutional length short-term memory neural network model, and the accuracy and continuity problems in polar sea ice density inversion are solved, and the sea ice coverage and time-varying are realized. Feature exploration.
Patent Information
- Application Number
- CN202411546828.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-01
- Publication Date
- 2024-12-27
- Estimated Expiration
- 2044-11-01
AI Technical Summary
The prior art has problems such as low accuracy, inconsistent results at continuous time points, unstable and noise data cannot be used as reference data in polar sea ice density inversion.
Convolutional length short-term memory neural network (ConvLSTM) model is used to obtain GNSS-R DDM data and sea ice density field data, perform bilinear interpolation and preprocessing, and build DDM-SIC sequence data sets, and use the trained The neural network model performs sea ice density inversion, and selects the inversion value of the middle position of the sequence for average calculation to obtain the final result.
It improves the accuracy and continuity of polar sea ice density inversion, solves the problem of noise data usage, and realizes effective exploration of sea ice coverage and time-varying characteristics.
Smart Images
Figure CN119046347B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the technical field of polar sea ice data inversion, and in particular to a polar sea ice density inversion method and device. Background Art
[0002] Polar sea ice plays an important role in regulating the global climate system. Sea ice concentration (SIC) is one of the important parameters to describe the state of polar sea ice. It refers to the percentage of sea ice coverage per unit area. Global Navigation Satellite System Reflectometry (GNSS-R) is a new low-cost remote sensing technology with the advantages of small size, good quality, low power consumption and high spatial resolution. It is highly favored in sea ice satellite remote sensing research. The Delay Doppler Map Observation (DDM) obtained by GNSS-R technology can distinguish sea water and sea ice based on waveform characteristics and obtain sea ice concentration.
[0003] With the launch of the TechDemoSat-1 (TDS-1) satellite in 2014, millions of delay-Doppler map data were provided, which greatly accelerated the application of DDM in sea ice detection and classification. Yan et al. (2017) first demonstrated the feasibility of DDM based on neural network to retrieve SIC; subsequently, further research on SIC retrieval was conducted by establishing the relationship between DDM and SIC based on deep learning or statistical models.
[0004] Previous studies often used a single DDM data set to invert the sea ice density at the corresponding mirror point. This one-to-one mapping relationship lost the temporal variation pattern, resulting in low inversion accuracy and instability. At the same time, the noise data in the DDM sequence data cannot be used as the input for inversion, which reduces the applicability of the inversion method. Summary of the invention
[0005] In view of this, the present application aims to propose a polar sea ice density inversion method and device to solve the problems of low polar sea ice density inversion accuracy, incoherent and unstable sea ice density inversion results at consecutive time points, and at the same time solve the problem that noise data cannot be used as reference data for sea ice inversion.
[0006] To achieve the above purpose, the technical solution of this application is implemented as follows:
[0007] In a first aspect, the present application provides a polar sea ice density inversion method, comprising:
[0008] Obtain GNSS-R DDM data and the corresponding polar sea ice density field data;
[0009] According to the longitude and latitude positions of the mirror point and the sea ice density field data, the sea ice density data at the mirror point corresponding to the DDM data at the corresponding time is calculated by bilinear interpolation method, and the DDM-SIC sequence data is constructed according to the preprocessed sea ice density data and the DDM data, and the DDM-SIC data set is generated in a chronological order;
[0010] A sea ice density inversion system is constructed based on the established neural network model, and the DDM data is processed by the sea ice density inversion system to obtain the sea ice density inversion value at the corresponding mirror point, wherein the neural network model is a convolutional long short-term memory neural network model trained by the DDM-SIC data set;
[0011] Multiple sea ice density inversion values located in the middle of the sequence are selected, and the multiple sea ice density inversion values are averaged to obtain the final inversion result of sea ice density in the polar region.
[0012] In the second aspect, based on the same inventive concept, the present application also provides a polar sea ice density inversion device, comprising:
[0013] A data acquisition module is configured to acquire GNSS-R DDM data and polar region sea ice density field data at a corresponding time;
[0014] A data set construction module is configured to calculate the sea ice density data at the mirror point corresponding to the DDM data at the corresponding time by bilinear interpolation according to the longitude and latitude positions of the mirror point and the sea ice density field data, construct DDM-SIC sequence data according to the preprocessed sea ice density data and the DDM data, and generate a DDM-SIC data set in a chronological order;
[0015] an inversion system construction module, configured to construct a sea ice density inversion system based on the established neural network model, and to process the DDM data through the sea ice density inversion system to obtain the sea ice density inversion value at the corresponding mirror point, wherein the neural network model is a convolutional long short-term memory neural network model trained by the DDM-SIC data set;
[0016] The inversion result output module is configured to select multiple sea ice density inversion values located in the middle of the sequence, and average the multiple sea ice density inversion values to obtain the final inversion result of the sea ice density in the polar region.
[0017] In a third aspect, based on the same inventive concept, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect is implemented.
[0018] In a fourth aspect, based on the same inventive concept, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method described in the first aspect.
[0019] Compared with the prior art, the polar sea ice density inversion method and device described in this application have the following beneficial effects:
[0020] The polar sea ice density inversion method and device described in the present application are a data pre-processing method for sorting and grouping continuous-time DDM-SIC sequence data, and a convolutional long short-term memory neural network (ConvLSTM) is used to invert the polar sea ice density. This method can solve the problems of low polar sea ice density inversion accuracy, incoherent and unstable sea ice density inversion results at continuous time points, and at the same time solves the problem that noise data cannot be used as reference data for sea ice inversion, thereby realizing the exploration of sea ice coverage in polar regions, the inversion of sea ice density, and the acquisition of time-varying characteristics of sea ice. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:
[0022] Figure 1 This is a flow chart of the polar sea ice density inversion method described in an embodiment of the present application;
[0023] Figure 2 This is a schematic diagram of constructing a DDM-SIC sample set according to an embodiment of the present application;
[0024] Figure 3 This is a schematic diagram of the structure of a convolutional long short-term memory neural network described in an embodiment of the present application;
[0025] Figure 4 An application roadmap for the sea ice density inversion system described in the embodiment of the present application;
[0026] Figure 5 This is a schematic diagram of the structure of the polar sea ice density inversion system described in an embodiment of the present application;
[0027] Figure 6 A schematic diagram of the hardware structure of the electronic device described in an embodiment of the present application. DETAILED DESCRIPTION
[0028] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.
[0029] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be the usual meanings understood by people with ordinary skills in the field to which the present application belongs. The "first", "second" and similar words used in the embodiments of the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.
[0030] GNSS-R technology can be used to obtain surface information at the mirror point where the reflected signal is reflected, which is manifested in the delay-Doppler field as differences in waveform, signal power and related characteristics. This delay-Doppler signal data is usually called a DDM image. The surface of the seawater is relatively rough, and its reflected signal is mainly composed of incoherent components, with a high degree of diffusion, and the seawater DDM is horseshoe-shaped; the reflected signal of sea ice is mainly composed of coherent components, with concentrated energy, and the sea ice DDM is dot-shaped. This provides data support for the inversion of sea ice density using DDM data.
[0031] The accuracy of the sea ice density of a single mirror point inverted using a single DDM is difficult to guarantee, and is often affected by noise data. Considering the regularity of the movement of mirror points on the earth's surface and the continuous and gradual change of sea ice density, using a continuous-time DDM atlas as the sea ice density of the mirror point trajectory can effectively solve the problem of insufficient accuracy, and make up for the inability to invert caused by noise data through the sea ice density data information of the previous and next time points.
[0032] ConvLSTM combines the convolutional neural network's ability to extract image feature information with the long short-term memory neural network's ability to obtain contextual information of time-varying sequences, and performs inversion and prediction based on continuous two-dimensional data.
[0033] The embodiments of the present application are described in detail below with reference to the accompanying drawings.
[0034] See also Figure 1As shown, this embodiment provides a polar sea ice density inversion method, which specifically includes the following steps:
[0035] Step S101, obtaining GNSS-R DDM data and polar region sea ice density field data at a corresponding time.
[0036] Specifically, in this embodiment, a large amount of DDM data (DDM data have the same shape, the size of the delay dimension is delaysize, and the size of the Doppler dimension is dopplersize) and the sea ice density field data in the polar region at the corresponding time are obtained based on the GNSS-R data of Yunyao Aerospace.
[0037] Step S102: According to the longitude and latitude positions of the mirror point and the sea ice density field data, the sea ice density data at the mirror point corresponding to the DDM data at the corresponding time is calculated by bilinear interpolation, and the DDM-SIC sequence data is constructed according to the preprocessed sea ice density data and DDM data, and the DDM-SIC data set is generated in chronological order.
[0038] Specifically, in this embodiment, by analyzing the delay dimension and Doppler dimension information in the DDM data and combining the orbit information of the satellite and the receiver, the longitude and latitude positions of the mirror point when the signal is reflected on the earth's surface are derived.
[0039] The latitude and longitude coordinates of the mirror point ( , ), that is, the geographical location of the specular reflection point, sea ice density field data: This data is usually a two-dimensional matrix, which represents the distribution of sea ice density at different longitude and latitude grid points. Assume that the longitude and latitude of the grid point is ( , ), the corresponding sea ice density is .
[0040] Find the longitude of the mirror point and latitude The grid unit where it is located, that is:
[0041] Find the four grid points closest to the mirror point ( , )、( , )、( , ) and( , ).
[0042] The bilinear interpolation method uses the values of these four adjacent grid points to estimate the sea ice density at the mirror point. The interpolation formula is as follows:
[0043] ;
[0044] In the formula, is the sea ice density of four adjacent grid points in the sea ice density field; , They are the longitude grids on both sides of the mirror point; , They are the latitude grids on both sides of the mirror point;
[0045] The longitude and latitude coordinates of the mirror point ( , ) and the density data of the corresponding grid points Substitute the bilinear interpolation formula to calculate the sea ice density at the mirror point .
[0046] The DDM data are preprocessed, including screening and normalization operations, to remove erroneous data and noise data for a long period of time, and the DDM data and sea ice density data are normalized to between (0, 1) to facilitate faster convergence of the neural network model.
[0047] The DDM-SIC sequence data are constructed based on the preprocessed sea ice density data and DDM data, and the DDM-SIC dataset is generated in chronological order.
[0048] In some embodiments, generating a DDM-SIC dataset in a chronological order includes:
[0049] intercepting DDM-SIC sequence data with a continuous time length, sorting and grouping the DDM-SIC sequence data in time order to form a plurality of sample sets in a rolling manner, wherein the sample sets retain noise data;
[0050] The DDM-SIC dataset consists of multiple sample sets.
[0051] Specifically, in this embodiment, a DDM-SIC sample set is generated according to the data set of datasize that is sorted out, and the DDM-SIC sequence data is arranged in chronological order. The specific process of generating the sample set is as follows: Figure 2 shown.
[0052] Determine the time length T of a single input atlas to ensure that there are enough valid DDM data within the time length T, and the threshold range of sea ice density is large enough. At the same time, the value of T should not be too large to avoid excessive training and use costs.
[0053] Sample sets are generated by grouping in sequence according to the time length T determined by a single sample set. The shape of the input DDM sample set X is (datasize, T, delaysize, dopplersize) and the shape of the output SIC sample set Y is (datasize, T).
[0054] In the process of generating the sample set, the noise data appearing in the time series are selected and retained (the noise data here is different from the noise data in the preprocessing process. The noise data eliminated in the preprocessing process is the noise data that appears continuously for a long time, and the noise data here is the data of a short time, so as to ensure the continuity of the data in time) to maintain the continuity of the data in time and realize the inversion of the sea ice density at the mirror point corresponding to the noise data.
[0055] Step S103, constructing a sea ice density inversion system based on the established neural network model, and processing the DDM data through the sea ice density inversion system to obtain the sea ice density inversion value at the corresponding mirror point, wherein the neural network model is a convolutional long short-term memory neural network model trained by the DDM-SIC data set.
[0056] Specifically, in this embodiment, the order of the data set is shuffled to ensure its wide applicability, and the data set is split into a training set and a validation set at a ratio of 7:3. A convolutional long short-term memory neural network model is constructed, and the mapping relationship between input X and output Y is learned using the training set, and the validation set is used for comparison, and the parameters in the neural network are gradually optimized. The structure of the convolutional long short-term memory neural network is as follows: Figure 3 shown.
[0057] The ConvLSTM layer extracts the spatial feature information of the DDM data sequence through convolution operations and uses the gating system to regulate the temporal variation information;
[0058] BatchNomalization layer, which performs normalization to speed up convergence;
[0059] Construct three ConvLSTM layer-BatchNormalization layer combinations to extract relevant features and important information multiple times;
[0060] Dense layer, fully connected layer, used for connection and parameter optimization of input feature values and output values;
[0061] Dropout layer, which is used to prevent overfitting;
[0062] Construct two consecutive Dense layers-Dropout layers to finally get the output value.
[0063] Adjust the parameters such as ConvLSTM units, number of neurons in the fully connected layer, Dropout rate, number of epochs, etc. in the neural network to build the optimal sea ice density inversion system.
[0064] The optimal sea ice density inversion system is applied multiple times to obtain the sea ice density value at the target mirror point. For each DDM data, it can be placed at any position in the atlas sequence, so T inversion values can be obtained.
[0065] Step S104: select a plurality of sea ice density inversion values located in the middle of the sequence, and average the plurality of sea ice density inversion values to obtain a final inversion result of the sea ice density in the polar region.
[0066] Specifically, in this embodiment, considering that the inversion capability at the middle position in the sequence is optimal, multiple inversion values can be averaged to obtain the final inversion result of the sea ice density in the polar region.
[0067] Figure 4 The data inversion roadmap for applying the above-mentioned sea ice concentration inversion system is as follows:
[0068] (1) Packing the DDM data, intercepting T consecutive DDM data to form a data sequence, and rolling to form multiple data sequences;
[0069] (2) The constructed sea ice density inversion system is used to process the DDM data series to obtain the corresponding sea ice density inversion value at the mirror point;
[0070] (3) Arrange the inverted values of sea ice density in sequence, and perform arithmetic averaging on the inverted values corresponding to the same mirror point but at different sequence positions to obtain the final value.
[0071] The polar sea ice density inversion method described in this embodiment is a data pre-processing method for sorting and grouping continuous time DDM-SIC sequence data, and adopts a convolutional long short-term memory neural network (ConvLSTM) to invert the polar sea ice density. This method can solve the problems of low polar sea ice density inversion accuracy, incoherent and unstable sea ice density inversion results at continuous time points, and at the same time solves the problem that noise data cannot be used as reference data for sea ice inversion, thereby realizing the exploration of sea ice coverage in polar regions, the inversion of sea ice density, and the acquisition of time-varying characteristics of sea ice.
[0072] It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0073] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, an embodiment of the present application also provides a polar sea ice density inversion device.
[0074] like Figure 5 As shown, the polar sea ice density inversion device comprises:
[0075] The data acquisition module 11 is configured to acquire GNSS-R DDM data and polar region sea ice density field data at a corresponding time;
[0076] The data set construction module 12 is configured to calculate the sea ice density data at the mirror point corresponding to the DDM data at the corresponding time by bilinear interpolation according to the latitude and longitude positions of the mirror point and the sea ice density field data, construct DDM-SIC sequence data according to the pre-processed sea ice density data and DDM data, and generate a DDM-SIC data set in a chronological order;
[0077] The inversion system construction module 13 is configured to construct a sea ice density inversion system based on the established neural network model, and process the DDM data through the sea ice density inversion system to obtain the sea ice density inversion value at the corresponding mirror point, wherein the neural network model is a convolutional long short-term memory neural network model trained by the DDM-SIC data set;
[0078] The inversion result output module 14 is configured to select a plurality of sea ice density inversion values located in the middle of the sequence, and average the plurality of sea ice density inversion values to obtain a final inversion result of the sea ice density in the polar region.
[0079] For the convenience of description, the above devices are described in terms of functions and are divided into various modules. Of course, when implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware.
[0080] The device of the above embodiment is used to implement the corresponding method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.
[0081] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, an embodiment of the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in any of the above embodiments is implemented.
[0082] Figure 6 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.
[0083] The processor 1010 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.
[0084] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory 1020 and are called and executed by the processor 1010.
[0085] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.
[0086] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).
[0087] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010 , the memory 1020 , the input / output interface 1030 , and the communication interface 1040 ).
[0088] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiments of the present specification, and does not necessarily include all the components shown in the figure.
[0089] The electronic device of the above embodiment is used to implement the corresponding method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.
[0090] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method described in any of the above embodiments.
[0091] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.
[0092] The computer instructions stored in the storage medium of the above embodiments are used to enable the computer to execute the method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.
[0093] A person skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. In line with the concept of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.
[0094] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power / ground connections to the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device may be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present application are to be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). Where specific details (e.g., circuits) are set forth to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or with changes in these specific details. Therefore, these descriptions should be considered illustrative rather than restrictive.
[0095] Although the present application has been described in conjunction with specific embodiments of the present application, many alternatives, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the discussed embodiments.
[0096] The embodiments of the present application are intended to cover all such substitutions, modifications and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of the present application.
Claims
1. A polar sea ice density inversion method, characterized in that: include: Obtain GNSS-R DDM data and the corresponding polar sea ice density field data; According to the longitude and latitude positions of the mirror point and the sea ice density field data, the sea ice density data at the mirror point corresponding to the DDM data at the corresponding time is calculated by bilinear interpolation method, and the DDM-SIC sequence data is constructed according to the preprocessed sea ice density data and the DDM data, and the DDM-SIC data set is generated in a chronological order; A sea ice density inversion system is constructed based on the established neural network model, and the DDM data is processed by the sea ice density inversion system to obtain the sea ice density inversion value at the corresponding mirror point, wherein the neural network model is a convolutional long short-term memory neural network model trained by the DDM-SIC data set; Multiple sea ice density inversion values located in the middle of the sequence are selected, and the multiple sea ice density inversion values are averaged to obtain the final inversion result of sea ice density in the polar region.
2. The method according to claim 1, characterized in that: The preprocessing process includes: The DDM data are screened to eliminate erroneous data and noise data for a long period of time, and the sea ice density data and the DDM data after the elimination are normalized.
3. The method according to claim 1, characterized in that The generating of the DDM-SIC data set in a chronological order includes: intercepting DDM-SIC sequence data with a continuous time length, sorting and grouping the DDM-SIC sequence data in time order to form a plurality of sample sets in a rolling manner, wherein the sample sets retain noise data; The DDM-SIC dataset consists of multiple sample sets.
4. The method according to claim 1, characterized in that: The DDM-SIC dataset is shuffled in order, and the dataset is split into a training set and a validation set in a ratio of 7:3; The training set is used to perform model training on the neural network model, and the verification set is used to perform model verification on the neural network model.
5. The method according to claim 1, characterized in that The neural network model structure includes: A ConvLSTM layer-BatchNormalization layer combination and a Dense layer-Dropout layer combination connected in sequence, wherein the ConvLSTM layer-BatchNormalization layer combination is composed of a plurality of alternately arranged ConvLSTM layers and BatchNormalization layers, wherein the first ConvLSTM layer is used to receive input data; The Dense layer-Dropout layer combination consists of a plurality of alternately arranged Dense layers and Dropout layers, wherein the last Dropout layer is used to output data.
6. A polar sea ice density inversion device, characterized in that: include: A data acquisition module is configured to acquire GNSS-R DDM data and polar region sea ice density field data at a corresponding time; A data set construction module is configured to calculate the sea ice density data at the mirror point corresponding to the DDM data at the corresponding time by bilinear interpolation according to the longitude and latitude positions of the mirror point and the sea ice density field data, construct DDM-SIC sequence data according to the preprocessed sea ice density data and the DDM data, and generate a DDM-SIC data set in a chronological order; an inversion system construction module, configured to construct a sea ice density inversion system based on the established neural network model, and to process the DDM data through the sea ice density inversion system to obtain the sea ice density inversion value at the corresponding mirror point, wherein the neural network model is a convolutional long short-term memory neural network model trained by the DDM-SIC data set; The inversion result output module is configured to select multiple sea ice density inversion values located in the middle of the sequence, and average the multiple sea ice density inversion values to obtain the final inversion result of the sea ice density in the polar region.
7. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 5 is implemented.
8. A non-transitory computer-readable storage medium, characterized in that: in, The non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions are used to cause a computer to execute the method according to any one of claims 1 to 5.