A method and system for offshore short-term sea surface variable prediction

Abstract

This invention provides a method and system for short-term sea surface variable prediction in nearshore waters. The method includes: forming regularized time-series data from drifting buoy observation data acquired within a preset nearshore area; dividing the regularized time-series data into training samples and comparison samples; extracting comparison target labels from the comparison samples; training an initial model using features from the training samples to obtain predicted target labels; obtaining a sea surface variable prediction model when the loss function value between the predicted target label and the comparison target label is less than a preset threshold; inputting the time-series observation data features of drifting buoys in nearshore waters before the prediction start time into the sea surface variable prediction model; and outputting the prediction results of sea surface variables for a preset time period after the prediction start time. Based on drifting buoy trajectory observation data, this method enables rapid short-term prediction of sea surface variables in China's nearshore waters, offering advantages such as strong scalability and ease of engineering deployment.

Description

Technical Field

[0001] This invention relates to the field of marine environmental forecasting, specifically to a method and system for predicting short-term sea surface variables in nearshore areas. Background Technology

[0002] China's coastal waters encompass diverse marine dynamic environments, including marginal seas, shelf seas, and straits. Sea surface temperature and surface current velocity are crucial for ship navigation, offshore operations, emergency response, and drift path determination. Existing short-term forecasting schemes typically rely on complex numerical models or external analysis fields, which suffer from high deployment costs, insufficient interpretability, and poor adaptability to in-situ continuous observations.

[0003] However, existing technologies lack a unified short-term forecasting scheme for China's coastal scenarios, especially a target-configurable forecasting method and system that can adapt to both sea surface temperature and surface velocity components. Summary of the Invention

[0004] This invention provides a nearshore short-term sea surface variable prediction system that can solve at least one technical problem in the prior art.

[0005] To achieve the above objectives, in one aspect, embodiments of the present invention provide a method for predicting short-term sea surface variables in nearshore areas, comprising:

[0006] The drifting buoy observation data obtained within a preset nearshore waters will be formed into regularized time-series data. The regularized time-series data will characterize the current state, historical evolution, and positional movement of the drifting buoy. The regularized time-series data will be divided into training samples and comparison samples. Comparison target labels will be extracted from the comparison samples. The target labels refer to the sea surface variable values ​​at multiple future moments within a preset future time period that correspond to the features of the training samples.

[0007] The initial model is trained using the features of the training samples to obtain the predicted target label; the training of the initial model is completed when the loss function value between the predicted target label and the compared target label is less than a preset threshold, and the sea surface variable prediction model is obtained.

[0008] When making short-term predictions of sea surface variables in a specified nearshore area, the time-series observation data characteristics of the drifting buoys in the nearshore area before the prediction start time are input into the sea surface variable prediction model, and the prediction results of the sea surface variables for a future preset time after the prediction start time are output.

[0009] On the other hand, embodiments of the present invention provide a nearshore short-term sea surface variable prediction system, comprising:

[0010] The sample construction unit is used to form regularized time-series data from drifting buoy observation data acquired within a preset nearshore waters. The regularized time-series data characterizes the current state, historical evolution, and positional movement of the drifting buoy. The regularized time-series data is divided into training samples and comparison samples. Comparison target labels are extracted from the comparison samples. The target labels refer to the sea surface variable values ​​at multiple future moments within a preset future time period that correspond to the features of the training samples.

[0011] The training unit is used to train the initial model using the features of the training samples to obtain the predicted target label; the training of the initial model is completed when the loss function value between the predicted target label and the compared target label is less than a preset threshold, and the sea surface variable prediction model is obtained.

[0012] The prediction unit is used to input the time-series observation data characteristics of the drifting buoys in the specified nearshore waters before the prediction start time into the sea surface variable prediction model when making short-term predictions of sea surface variables in the specified nearshore waters, and output the prediction results of the sea surface variables for a future preset time after the prediction start time.

[0013] The above technical solution offers the following advantages: Drifting buoys can continuously provide information on sea surface temperature, geographical location, and surface motion, offering advantages such as high observation frequency, continuous trajectories, and convenient data acquisition. Based on drifting buoy trajectory observation data, short-term rapid predictions of sea surface variables in China's coastal waters are possible, offering advantages such as strong scalability and ease of engineering deployment. By standardizing drifting buoy observations, constructing multi-source time-series features, and establishing a sea surface variable prediction model, forward-looking predictions of sea surface variables for a predetermined timeframe can be achieved.

[0014] The system revolves around a core technology chain encompassing data acquisition, time warping, feature construction, target alignment, and model prediction. It employs a unified data processing chain compatible with different sea surface variables, facilitating switching between sea surface temperature and surface velocity. By incorporating historical evolution, positional motion, and temporal context information, it corrects purely continuous predictions, enhancing short-term forecast stability. The system modules have clear boundaries, facilitating deployment within marine operational platforms. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a method for predicting short-term sea surface variables in nearshore areas according to an embodiment of the present invention;

[0017] Figure 2 This is a schematic diagram of the structure of a nearshore short-time sea surface variable prediction system according to an embodiment of the present invention;

[0018] Figure 3 This is a flowchart illustrating a method for predicting short-term sea surface variables in nearshore waters according to an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] like Figure 1 As shown, in conjunction with embodiments of the present invention, a method for predicting short-term sea surface variables in nearshore areas is provided, comprising:

[0021] S101: The drifting buoy observation data obtained within the preset nearshore waters will be formed into regularized time-series data. The regularized time-series data will characterize the current state, historical evolution, and positional movement of the drifting buoy. The regularized time-series data will be divided into training samples and comparison samples. Comparison target labels will be extracted from the comparison samples. The target labels refer to the sea surface variable values ​​at multiple future moments within a preset future time period that correspond to the features of the training samples.

[0022] S102: Train the initial model using the features of the training samples to obtain the predicted target label; the training of the initial model is completed when the loss function value between the predicted target label and the compared target label is less than the preset threshold, and the sea surface variable prediction model is obtained.

[0023] S103: When making short-term predictions of sea surface variables in a specified nearshore area, the time-series observation data characteristics of the drifting buoys in the nearshore area before the prediction start time are input into the sea surface variable prediction model, and the prediction results of the sea surface variables for a future preset duration after the prediction start time are output.

[0024] "Short time" is a relative concept, and the generally accepted understanding usually refers to a time scale of a few hours to a few days.

[0025] Training sample features (features constructed from training samples) are a set of variables used as input to the initial model to characterize the current state and its evolution trend.

[0026] The current state is characterized by the current observed values ​​of sea surface variables, while historical evolution refers to the changing trends and patterns of sea surface variables over a period of time. Positional motion includes the geographical location characteristics (latitude and longitude) and surface motion characteristics (east-west / north-south direction) of drifting buoys.

[0027] Based on drifting buoy trajectory observation data, it is possible to make short-term and rapid predictions of sea surface variables in China's coastal waters, with the advantages of strong scalability and ease of engineering deployment. By standardizing drifting buoy observations, constructing multi-source time-series features, and establishing a sea surface variable prediction model, forward predictions of sea surface variables for a predetermined time period can be achieved.

[0028] Preferably, in S101, the nearshore short-term sea surface variable prediction method includes:

[0029] S101-1: Group the drifting buoy observation data acquired within a preset nearshore waters area according to the platform identifier, sort the data in chronological order within each group, and perform at least one operation of resampling, aggregation, alignment, or interpolation on the drifting buoy observation data according to a preset time granularity, i.e., time context, to obtain time-normalized observation data, thereby realizing time normalization of drifting buoy observation data. The drifting buoy observation data mainly includes sea surface temperature, latitude, longitude, east-west surface velocity, north-south surface velocity, timestamp, and platform identifier. The source of drifting buoy observation data can be a public or private drifting buoy data source.

[0030] S101-2: The time-warped observation data is screened for missing data and anomalies are removed, and consistency is verified to obtain regularized time-series data, thereby achieving quality control.

[0031] The time context includes not only timestamps, but also time indices (such as days within the year and hours within the day) can be processed by sine / cosine transformations to obtain time periodic features that can characterize the periodic patterns of days, years, etc.

[0032] Preferably, dividing the regularized time-series data into training samples and comparison samples further includes:

[0033] S104: For the regularized time series data of the same drifting buoy, the training sample features are constructed using historical observations before a selected time, and the sea surface variable observations within a preset time period after the selected time are used as the corresponding target labels. The two are aligned by time to obtain training samples and comparison samples respectively.

[0034] A trajectory is a data sequence consisting of all time-series observations from the same platform (i.e., the same drifting buoy).

[0035] When constructing target labels, target sea surface variables on the same platform trajectory are shifted and aligned forward according to a preset future time interval. This establishes a sample correspondence between the training sample features constructed at the selected time and the target label values ​​at future times. The application of forward alignment to the trajectory is as follows: within the time series of the same buoy, for a certain time (T), the target label value (e.g., sea surface temperature) for the corresponding future time (e.g., T+24 hours) is "shifted" back on the time axis according to the preset future time interval, serving as the target label to be learned by the training sample features at time T. This process is completed entirely on the trajectory of the same drifting buoy, ensuring the continuity of the physical attributes of the predicted object and avoiding the erroneous use of the future value of buoy A as the label of buoy B. The preset future time interval can be 24 hours or other durations as needed.

[0036] Preferably, the current state is characterized by the current observed values ​​of sea surface variables, which include at least one of the following: sea surface temperature, east-west surface velocity, and north-south surface velocity.

[0037] During prediction, the target label can be switched between different sea surface variables without changing the basic processing flow, namely S101 and S102. Because the training sample feature construction stage includes different sea surface variables (sea surface temperature, east-west surface velocity, and north-south surface velocity), when making short-term predictions of sea surface variables in a specified nearshore area for different target labels, such as switching from predicting sea surface temperature to predicting east-west surface velocity, only the direction of the target label needs to be changed. There is no need to reacquire new drifting buoy observation data to form regularized time-series data for training the initial model.

[0038] Preferably, the time-series observation data characteristics of the nearshore drifting buoy include at least one of the following: current observation characteristics, lag time characteristics, rolling statistics characteristics, geographical location characteristics, surface motion characteristics, and time period characteristics of the drifting buoy, wherein:

[0039] The lag time features include sea surface variable features of drifting buoys corresponding to a historical moment that predicts the start time, or multiple adjacent historical moments; for example, the lag time features include features corresponding to at least one historical moment from the previous 1 hour, 3 hours, 6 hours and 24 hours.

[0040] The rolling statistical features include at least one of the mean, variance, or extreme values ​​of the sea surface variables of the drifting buoy, calculated based on a preset time window prior to the prediction start time.

[0041] The time period characteristic refers to the periodic mapping of sea surface variables of drifting buoys prior to the measurement start time according to a set index period. This periodic mapping includes sine or cosine transformations. The set index period can be, for example, based on the day within a year, the hour within a day, or other time index periods.

[0042] Geographical location features refer to the spatial coordinates of the drifting buoy, specifically determined by latitude and longitude.

[0043] Surface motion characteristics refer to the motion state information of drifting buoys, specifically determined by east-west surface velocity and north-south surface velocity.

[0044] Preferably, the initial model is one of the following: a linear regression model, a linear regression model with regularization, a tree model, and a neural network model; the linear regression model with regularization is a ridge regression model with L2 regularization.

[0045] When constructing training samples, samples that simultaneously meet the preset feature integrity condition and the target label availability condition are retained, ensuring consistency of training samples across feature dimensions. The preset feature integrity condition means that all feature values ​​(such as current observation, lag features, location features, etc.) constituting the training sample must exist and cannot be missing or empty labels. This is directly guaranteed by the missing data screening and anomaly removal in step S101-2 corresponding to the regularized time-series data.

[0046] The condition for target label availability means that the target label corresponding to the sample must be real and valid. Specifically, when extracting target label values ​​as labels from future time points using the forward alignment method, the observation at that future time point must actually exist. This excludes historical records where no labels are available due to possible reasons such as buoy signal interruption.

[0047] like Figure 2 As shown, in conjunction with embodiments of the present invention, a nearshore short-term sea surface variable prediction system is provided, comprising:

[0048] The sample construction unit 21 is used to form regularized time series data from drifting buoy observation data acquired within a preset nearshore waters. The regularized time series data characterizes the current state, historical evolution, and positional movement of the drifting buoy. The regularized time series data is divided into training samples and comparison samples. The comparison target label is extracted from the comparison sample. The target label refers to the sea surface variable value at multiple future moments within a preset future time period corresponding to the features of the training sample.

[0049] Training unit 22 is used to train the initial model using the features of training samples to obtain the predicted target label; the training of the initial model is completed when the loss function value between the predicted target label and the compared target label is less than a preset threshold, and the sea surface variable prediction model is obtained.

[0050] The prediction unit 23 is used to input the time series observation data characteristics of the drifting buoys in the specified nearshore waters before the prediction start time into the sea surface variable prediction model when making short-term predictions of sea surface variables in the specified nearshore waters, and output the prediction results of the sea surface variables for a future preset time after the prediction start time.

[0051] "Short time" is a relative concept, and the generally accepted understanding usually refers to a time scale of a few hours to a few days.

[0052] Training sample features (features constructed from training samples) are a set of variables used as input to the initial model to characterize the current state and its evolution trend.

[0053] The current state is characterized by the current observed values ​​of sea surface variables, while historical evolution refers to the changing trends and patterns of sea surface variables over a period of time. Positional motion includes the geographical location characteristics (latitude and longitude) and surface motion characteristics (east-west / north-south direction) of drifting buoys.

[0054] Based on drifting buoy trajectory observation data, it is possible to make short-term and rapid predictions of sea surface variables in China's coastal waters, with the advantages of strong scalability and ease of engineering deployment. By standardizing drifting buoy observations, constructing multi-source time-series features, and establishing a sea surface variable prediction model, forward predictions of sea surface variables for a predetermined time period can be achieved.

[0055] Preferably, the nearshore short-time sea surface variable prediction system further includes a preprocessing unit for:

[0056] According to the platform identifier, the drifting buoy observation data acquired within a preset nearshore waters area is grouped according to the platform identifier and sorted chronologically within each group. The drifting buoy observation data is then subjected to at least one of the following operations: resampling, aggregation, alignment, or interpolation, according to a preset time granularity, i.e., time context, to obtain time-normalized observation data. This achieves time normalization of the drifting buoy observation data. The drifting buoy observation data mainly includes sea surface temperature, latitude, longitude, east-west surface velocity, north-south surface velocity, timestamp, and platform identifier. The source of the drifting buoy observation data can be a public or private drifting buoy data source.

[0057] The time-warped observation data is subjected to missing data screening and anomaly removal, and consistency verification is performed to obtain regularized time-series data, thereby achieving quality control.

[0058] The time context includes not only timestamps, but also time periodic features that can characterize the periodic patterns of days and years by processing time indices (such as days within the year and hours within the day) through sine / cosine transformation.

[0059] Preferably, the sample construction unit 21 is specifically used for:

[0060] For the regularized time-series data of the same drifting buoy, the features of the training samples are constructed using historical observations before a selected time, and the sea surface variable observations within a preset time period after the selected time are used as the corresponding target labels. The two are aligned in time to obtain the training samples and comparison samples respectively.

[0061] A trajectory is a data sequence consisting of all time-series observations from the same platform (i.e., the same drifting buoy).

[0062] When constructing target labels, target sea surface variables on the same platform trajectory are shifted forward and aligned according to a preset future time interval. This establishes a sample correspondence between the training sample features constructed at the selected time and the target label values ​​at future times. The application of forward alignment to the trajectory is as follows: within the time series of the same buoy, for a certain time (T), the target label value (e.g., sea surface temperature) at a future time (e.g., T+24 hours) is "shifted" back onto the time axis according to the preset future time interval, serving as the target label to be learned by the training sample features at time T. This process is completed entirely on the trajectory of the same drifting buoy, ensuring the continuity of the physical attributes of the predicted object and avoiding the erroneous use of the future value of buoy A as the label of buoy B. The preset future time interval can be 24 hours or other durations depending on operational needs.

[0063] Preferably, the current state is characterized by the current observed values ​​of sea surface variables, which include at least one of the following: sea surface temperature, east-west surface velocity, and north-south surface velocity.

[0064] During prediction, the target label can be switched between different sea surface variables without changing the basic processing flow, namely S101 and S102. Because the training sample feature construction stage includes different sea surface variables (sea surface temperature, east-west surface velocity, and north-south surface velocity), when making short-term predictions of sea surface variables in a specified nearshore area for different target labels, such as switching from predicting sea surface temperature to predicting east-west surface velocity, only the direction of the target label needs to be changed. There is no need to reacquire new drifting buoy observation data to form regularized time-series data for training the initial model.

[0065] Preferably, the time-series observation data characteristics of the nearshore drifting buoy include at least one of the following: current observation characteristics, lag time characteristics, rolling statistics characteristics, geographical location characteristics, surface motion characteristics, and time period characteristics of the drifting buoy, wherein:

[0066] The lag time features include sea surface variable features of drifting buoys corresponding to a historical moment that predicts the start time, or multiple adjacent historical moments; for example, the lag time features include features corresponding to at least one historical moment from the previous 1 hour, 3 hours, 6 hours and 24 hours.

[0067] The rolling statistical features include at least one of the mean, variance, or extreme values ​​of the sea surface variables of the drifting buoy, calculated based on a preset time window prior to the prediction start time.

[0068] The time period characteristic refers to the periodic mapping of sea surface variables of drifting buoys prior to the measurement start time according to a set index period. This periodic mapping includes sine or cosine transformations. The set index period can be, for example, based on the day within a year, the hour within a day, or other time index periods.

[0069] Geographical location features refer to the spatial coordinates of the drifting buoy, specifically determined by latitude and longitude.

[0070] Surface motion characteristics refer to the motion state information of drifting buoys, specifically determined by east-west surface velocity and north-south surface velocity.

[0071] Preferably, the initial model is one of the following: a linear regression model, a linear regression model with regularization, a tree model, and a neural network model; the linear regression model with regularization is a ridge regression model with L2 regularization.

[0072] Preferably, when constructing training samples, samples that simultaneously meet the preset feature integrity condition and the target label availability condition are retained, ensuring consistency of training samples across feature dimensions. The preset feature integrity condition means that all feature values ​​(such as current observation, lag features, location features, etc.) constituting the training sample must exist and cannot be missing or empty labels. This is directly guaranteed by the missing data screening and anomaly removal in step S101-2 corresponding to the regularized time-series data.

[0073] The condition for target label availability means that the target label corresponding to the sample must be real and valid. Specifically, when extracting target label values ​​as labels from future time points using the forward alignment method, the observation at that future time point must actually exist. This excludes historical records where no labels are available due to possible reasons such as buoy signal interruption.

[0074] The embodiments of the present invention have at least the following beneficial effects:

[0075] Drifting buoys continuously provide information on sea surface temperature, geographical location, and surface motion, offering advantages such as high observation frequency, continuous trajectories, and convenient data acquisition. Based on drifting buoy trajectory observation data, short-term rapid predictions of sea surface variables in China's coastal waters are possible, offering advantages such as strong scalability and ease of engineering deployment. By standardizing drifting buoy observations, constructing multi-source time-series features, and establishing sea surface variable prediction models, forward-looking predictions of sea surface variables for a predetermined timeframe can be achieved.

[0076] The system revolves around a core technology chain encompassing data acquisition, time warping, feature construction, target alignment, and model prediction. It employs a unified data processing chain compatible with different sea surface variables, facilitating switching between sea surface temperature and surface velocity. By incorporating historical evolution, positional motion, and temporal context information, it corrects purely continuous predictions, enhancing short-term forecast stability. The system modules have clear boundaries, facilitating deployment within marine operational platforms.

[0077] The technical solutions of the present invention will be described in detail below with reference to specific application examples. For technical details not described in the implementation process, please refer to the relevant descriptions above.

[0078] like Figure 3 As shown, the following variable symbols will be used uniformly: This represents the input feature vector; This represents the predicted value corresponding to a preset future duration; Indicates the target label; Represents the model parameter vector; Indicates the bias term; Indicates the total number of training samples; Represents the regularization coefficient; Indicates the number of hours within a day; Indicates the date within the year; Indicates the length of the cycle within the year; Indicates the characteristics of the current observation; Indicates the characteristics of time lag; Indicates rolling statistical characteristics; Indicates location features; Indicates motion characteristics; Indicates a time periodicity. Unless otherwise stated, the above symbols retain the same meaning throughout the text.

[0079] In one embodiment, the sea surface variable prediction model can be represented as an input feature vector. To the prediction results The mapping relationship can be generally expressed as:

[0080]

[0081] in, This represents the mapping function learned through training samples. Formula (1) is used to summarize the general expression of the sea surface variable prediction model in this invention.

[0082] When the sea surface variable prediction model adopts a linear model or a linear model with regularization terms, its prediction expression can be further expressed as:

[0083]

[0084] in, Represents the model parameter vector The transpose of . Equation (2) shows that the prediction result From the input feature vector With model parameter vector Linear combinations and bias terms To be determined jointly.

[0085] For the scenario described in this invention, the input feature vector It can be obtained by concatenating multiple types of features, and its schematic form can be represented as follows:

[0086]

[0087] in, Indicates the characteristics of the current observation; Indicates the characteristics of time lag; Indicates rolling statistical characteristics; Indicates location features; Indicates motion characteristics; This represents the time periodic characteristics. Formula (3) is the input feature vector. The schematic composition is used to illustrate the unified organizational structure of multiple types of input features in this invention.

[0088] In a preferred embodiment, the time period feature can be constructed according to the following formula: , , and .in, Indicates the number of hours within a day. Indicates the date within the year. This indicates the length of the cycle within a year. Through the above cycle mapping, the time characteristics can be kept continuous at the cycle boundaries.

[0089] In a preferred embodiment, the sea surface variable prediction model can employ a linear regression model with an L2 regularization term, i.e., a ridge regression model, whose objective function can be expressed as:

[0090]

[0091] in, Represent the objective function; Indicates the first The input feature vector of each training sample; Indicates the first The target label corresponding to each training sample; Represents the parameter vector The squared L2 norm. Formula (4) consists of a prediction error term and a regularization term, which can improve parameter stability and reduce the risk of overfitting while maintaining model simplicity.

[0092] In other implementations, other regression models or function mapping models that meet the purpose of short-term sea surface variable prediction may also be used.

[0093] in, Represent the objective function; Indicates the first The input feature vector of each training sample; Indicates the first The target label corresponding to each training sample; Represents the model parameter vector; Indicates the bias term; Represents the regularization coefficient; Represents the parameter vector The square of the L2 norm; This represents the total number of training samples. The objective function described above consists of a prediction error term and a regularization term, which can improve parameter stability and reduce the risk of overfitting while maintaining model simplicity.

[0094] In other implementations, other regression models or function mapping models that meet the purpose of short-term sea surface variable prediction may also be used.

[0095] Specifically, during model training, the first... Each training sample is represented as ,in, The input feature vector is composed of multiple types of input features. For the corresponding target label. This is achieved through the objective function. Optimization can yield the parameters of the sea surface variable prediction model. and bias terms The preferred ridge regression model retains the advantages of linear models in terms of ease of interpretation and deployment, while also improving parameter stability through regularization terms.

[0096] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process may be rearranged without departing from the scope of this disclosure. The appended method claims provide elements of various steps in an exemplary order and are not intended to limit the scope to the specific order or hierarchy described.

[0097] In the above detailed description, various features are combined together in a single embodiment to simplify this disclosure. This approach to disclosure should not be construed as reflecting an intention that embodiments of the claimed subject matter require more features than are explicitly stated in each claim. Rather, as reflected in the appended claims, the invention is presented with fewer features than all of the features of the single disclosed embodiment. Therefore, the appended claims are hereby explicitly incorporated into the detailed description, wherein each claim stands alone as a preferred embodiment of the invention.

[0098] The disclosed embodiments have been described above to enable any person skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be applied to other embodiments without departing from the spirit and scope of this disclosure. Therefore, this disclosure is not limited to the embodiments given herein, but is consistent with the broadest scope of the principles and novel features disclosed in this application.

[0099] The foregoing description includes examples of one or more embodiments. It is certainly impossible to describe all possible combinations of components or methods in order to describe the above embodiments, but those skilled in the art will recognize that further combinations and arrangements of the various embodiments are possible. Therefore, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of the appended claims. Furthermore, the term "comprising" as used in the specification or claims is interpreted in a manner similar to the term "including," as interpreted when used as a conjunction in the claims. Additionally, the use of any term "or" in the specification of the claims is intended to mean "non-exclusive or."

[0100] Those skilled in the art will also understand that the various illustrative logical blocks, units, and steps listed in the embodiments of the present invention can be implemented by electronic hardware, computer software, or a combination of both. To clearly demonstrate the interchangeability of hardware and software, the functions of the various illustrative components, units, and steps described above have been generally described. Whether such functionality is implemented through hardware or software depends on the specific application and the overall system design requirements. Those skilled in the art can implement the described functions using various methods for each specific application, but such implementation should not be construed as exceeding the scope of protection of the embodiments of the present invention.

[0101] The various illustrative logic blocks or units described in the embodiments of this invention can be implemented or operate the described functions using a general-purpose processor, digital signal processor, application-specific integrated circuit (ASIC), field-programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. The general-purpose processor can be a microprocessor; alternatively, it can be any conventional processor, controller, microcontroller, or state machine. The processor can also be implemented using a combination of computing devices, such as a digital signal processor and a microprocessor, multiple microprocessors, one or more microprocessors combined with a digital signal processor core, or any other similar configuration.

[0102] The steps of the methods or algorithms described in the embodiments of this invention can be directly embedded in hardware, a software module executed by a processor, or a combination of both. The software module can be stored in RAM, flash memory, ROM, EPROM, EEPROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium in the art. Exemplarily, the storage medium can be connected to the processor so that the processor can read information from and write information to the storage medium. Optionally, the storage medium can also be integrated into the processor. The processor and storage medium can be housed in an ASIC, which can be housed in a user terminal. Optionally, the processor and storage medium can also be housed in different components of the user terminal.

[0103] In one or more exemplary designs, the functions described in the embodiments of the present invention can be implemented in hardware, software, firmware, or any combination of these three. If implemented in software, these functions can be stored on a computer-readable medium or transmitted on a computer-readable medium in the form of one or more instructions or code. Computer-readable media include computer storage media and communication media that facilitate the transfer of computer programs from one place to another. Storage media can be any available media that can be accessed by a general-purpose or special-purpose computer. For example, such computer-readable media can include, but is not limited to, RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to carry or store program code in the form of instructions or data structures and other forms that can be read by a general-purpose or special-purpose computer, or a general-purpose or special-purpose processor. Furthermore, any connection can be suitably defined as a computer-readable medium, for example, if the software is transmitted from a website, server or other remote resource via a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL) or wirelessly, such as infrared, wireless and microwave, it is also included in the defined computer-readable medium. The disks and discs mentioned include compressed disks, laser discs, optical discs, DVDs, floppy disks, and Blu-ray discs. Disks typically copy data magnetically, while discs typically copy data optically using lasers. Combinations of the above can also be contained in computer-readable media.

[0104] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting short-term sea surface variables in nearshore waters, characterized in that, include: The drifting buoy observation data will be acquired within a preset nearshore waters to form regularized time-series data. The regularized time-series data will characterize the current state, historical evolution, and positional movement of the drifting buoy. The regularized time-series data will be divided into training samples and comparison samples. Comparison target labels will be extracted from the comparison samples. The target labels refer to the sea surface variable values ​​at multiple future moments within a preset future time period that correspond to the features of the training samples. The initial model is trained using the features of the training samples to obtain the predicted target label; the training of the initial model is completed when the loss function value between the predicted target label and the compared target label is less than a preset threshold, and the sea surface variable prediction model is obtained. When making short-term predictions of sea surface variables in a specified nearshore area, the time-series observation data characteristics of the drifting buoys in the nearshore area before the prediction start time are input into the sea surface variable prediction model, and the prediction results of the sea surface variables for a future preset time after the prediction start time are output.

2. The nearshore short-term sea surface variable prediction method according to claim 1, characterized in that, Also includes: According to the platform identifier, the drifting buoy observation data within the preset nearshore waters area is subjected to at least one of the following operations: resampling, aggregation, alignment, or interpolation, according to the preset time granularity, to obtain time-normalized observation data. The drifting buoy observation data includes sea surface temperature, latitude, longitude, east-west surface velocity, north-south surface velocity, timestamp, and platform identifier. The time-warped observation data is subjected to missing data screening and anomaly removal, and consistency verification is performed to obtain regularized time-series data.

3. The nearshore short-term sea surface variable prediction method according to claim 1, characterized in that, The regularized time-series data is divided into training samples and comparison samples, including: For regularized time-series data of the same drifting buoy, training sample features are constructed using historical observations before a selected time, and sea surface variable observations within a preset time period after the selected time are used as corresponding target labels, and aligned by time to obtain training samples and comparison samples respectively.

4. The method for predicting short-term sea surface variables in nearshore areas according to claim 1, characterized in that, The sea surface variables include at least one of the following: sea surface temperature, east-west surface velocity, and north-south surface velocity.

5. The method for predicting short-term sea surface variables in nearshore areas according to claim 1, characterized in that, The time-series observation data characteristics of the nearshore drifting buoy include at least one of the following: current observation characteristics, lag time characteristics, rolling statistical characteristics, geographical location characteristics, surface motion characteristics, and time period characteristics of the drifting buoy, wherein: The lag time features include the sea surface variable features of the drifting buoy corresponding to a historical moment that predicts the start time, or multiple adjacent historical moments. The rolling statistical features include at least one of the mean, variance, or extreme values ​​of the sea surface variables of the drifting buoy, calculated based on a preset time window prior to the prediction start time. The time period feature refers to the periodic mapping of the sea surface variables of the drifting buoy before the start time of the measurement according to the set index period. The periodic mapping includes sine transformation or cosine transformation.

6. The method for predicting short-term sea surface variables in nearshore areas according to claim 1, characterized in that, The initial model is one of the following: a linear regression model, a linear regression model with regularization, a tree model, and a neural network model; the linear regression model with regularization is a ridge regression model with L2 regularization.

7. A nearshore short-term sea surface variable prediction system, characterized in that, include: The sample construction unit is used to form regularized time-series data from drifting buoy observation data acquired within a preset nearshore waters. The regularized time-series data characterizes the current state, historical evolution, and positional movement of the drifting buoy. The regularized time-series data is divided into training samples and comparison samples. Comparison target labels are extracted from the comparison samples. The target labels refer to the sea surface variable values ​​at multiple future moments within a preset future time period that correspond to the features of the training samples. The training unit is used to train the initial model using the features of the training samples to obtain the predicted target label; the training of the initial model is completed when the loss function value between the predicted target label and the compared target label is less than a preset threshold, and the sea surface variable prediction model is obtained. The prediction unit is used to input the time-series observation data characteristics of the drifting buoys in the specified nearshore waters before the prediction start time into the sea surface variable prediction model when making short-term predictions of sea surface variables in the specified nearshore waters, and output the prediction results of the sea surface variables for a future preset time after the prediction start time.

8. The nearshore short-term sea surface variable prediction system according to claim 7, characterized in that, It also includes a preprocessing unit for: According to the platform identifier, short-term drifting buoy observation data within a preset nearshore waters area are subjected to at least one of the following operations: resampling, aggregation, alignment, or interpolation, according to a preset time granularity, to obtain time-normalized observation data. The drifting buoy observation data includes sea surface temperature, latitude, longitude, east-west surface velocity, north-south surface velocity, timestamp, and platform identifier. The time-warped observation data is subjected to missing data screening and anomaly removal, and consistency verification is performed to obtain regularized time-series data.

9. The nearshore short-term sea surface variable prediction system according to claim 7, characterized in that, Sample building unit, specifically used for: For regularized time-series data of the same drifting buoy, training sample features are constructed using historical observations before a selected time, and sea surface variable observations within a preset time period after the selected time are used as corresponding target labels, and aligned by time to obtain training samples and comparison samples respectively.

10. The nearshore short-term sea surface variable prediction system according to claim 7, characterized in that, The sea surface variables include at least one of the following: sea surface temperature, east-west surface velocity, and north-south surface velocity.

Images