Sea surface backscattering coefficient and Doppler velocity prediction method, device and equipment

Abstract

The invention discloses a sea surface backscattering coefficient and Doppler velocity prediction method, device and equipment, and belongs to the technical field of ocean remote sensing, and the method comprises the steps: randomly setting a plurality of groups of sea surface data; performing flow field simulation based on the sea surface data to obtain a plurality of groups of longitude and latitude flow fields; converting the longitude and latitude flow fields into radar flow field parameters; performing sea surface simulation based on each radar flow field parameter to generate a sea surface backscattering coefficient-Doppler velocity combination; combining the sea surface data, the radar flow field parameters and the sea surface backscattering coefficient-Doppler velocity combination to form a data pair; performing regression prediction training on the full-connection neural network by taking each data pair as a training sample, and forming a sea surface prediction model after the regression prediction training is completed; and using the sea surface prediction model to predict a sea surface backscattering coefficient prediction value and a Doppler velocity prediction value. Therefore, by implementing the method, the problem of low marine remote sensing parameter processing efficiency in the prior art can be solved.

Description

Technical Field

[0001] The present invention relates to the field of ocean remote sensing technology, in particular to a method, device and equipment for predicting sea surface backscatter coefficient and Doppler velocity. Background Art

[0002] In the field of ocean remote sensing, sea surface scattering and Doppler velocity are two key parameters crucial for understanding and predicting ocean dynamics, meteorological events, and climate change. The sea surface scattering coefficient can be used to infer ocean surface wind information from radar echoes, while the Doppler velocity can be used to infer ocean surface dynamics, providing precise measurements of flow velocity and direction for flow field research. Obtaining these parameters is crucial for marine resource development, environmental monitoring, climate research, and even military applications.

[0003] Traditional methods for obtaining sea surface backscatter coefficients and Doppler velocities primarily rely on physical and empirical models. Physical models can simulate the signal response of radar detecting the sea surface, but as the number of factors increases, the models become overly complex and unsuitable for rapid simulation. While empirical models offer high resolution, they are limited in their applicability to a single wavelength band, only yielding the average scattering coefficient for a specific region or the Doppler velocity for a specific point in time. These models are also incapable of adapting to rapidly changing ocean environments. Consequently, rapidly obtaining sea surface backscatter coefficients and Doppler velocities is difficult when considering multiple factors, including surface currents, wind fields, and radar information. Summary of the Invention

[0004] The present invention provides a method, device and equipment for predicting sea surface backscatter coefficient and Doppler velocity, which can solve the problem of low efficiency in processing ocean remote sensing parameters in the prior art.

[0005] In order to solve the above technical problems, the present invention provides a method for predicting sea surface backscatter coefficient and Doppler velocity, comprising:

[0006] Randomly setting a number of groups of sea surface data; wherein the sea surface data includes wind speed, radar polarization mode, radar band, radar incident angle and radar viewing angle;

[0007] Performing flow field simulation based on each of the sea surface data to obtain several groups of latitude and longitude flow fields; wherein each group of latitude and longitude flow fields includes a longitude flow field and a latitude flow field;

[0008] Converting each of the latitude and longitude flow fields into a plurality of groups of radar flow field parameters; wherein each group of radar flow field parameters includes a radar radial flow field and a radial flow field gradient;

[0009] Performing sea surface simulation based on each of the radar flow field parameters to generate a plurality of sea surface backscatter coefficient-Doppler velocity combinations;

[0010] Combining several sets of sea surface data, several sets of radar flow field parameters and several sea surface backscatter coefficient-Doppler velocity combinations according to corresponding relationships to form several data pairs;

[0011] Using each of the data pairs as training samples, performing regression prediction training on a fully connected neural network, and determining the fully connected neural network after the regression prediction training as a sea surface prediction model;

[0012] The real-time wind speed, real-time radar polarization mode, real-time radar band, real-time radar incident angle, real-time radar viewing angle, real-time radar radial flow field and real-time radial flow field gradient are input into the sea surface prediction model to obtain a sea surface backscatter coefficient prediction value and a Doppler velocity prediction value.

[0013] As a preferred solution, the flow field simulation is performed based on each of the sea surface data to obtain several sets of latitude and longitude flow fields, including:

[0014] For each sea surface data, the sea surface data is input into the Princeton Ocean Model, so that the Princeton Ocean Model performs simulation according to the wind speed, radar polarization mode, radar band, radar incidence angle and radar viewing angle in the sea surface data to obtain flow velocity data at several different sea surface positions, and then analyzes the flow velocity data at several different sea surface positions to obtain longitude flow field and latitude flow field, and then generates latitude and longitude flow field according to the longitude flow field and the latitude flow field.

[0015] As a preferred solution, the latitude and longitude flow fields are converted into several groups of radar flow field parameters, including:

[0016] Obtain the radar viewing angle corresponding to each longitude and latitude flow field respectively;

[0017] For each latitude and longitude flow field, the radar radial flow field is calculated using the following formula:

[0018] Vx1=cos(los)*Vx0+sin(los)*Vy0

[0019] Where Vx1 is the radar radial flow field; Vx0 is the latitude flow field; Vy0 is the longitude flow field; and los is the radar viewing angle.

[0020] As a preferred solution, the latitude and longitude flow fields are converted into several groups of radar flow field parameters, including:

[0021] For each latitude and longitude flow field, the radial flow field gradient is calculated using the following formula:

[0022]

[0023] Where, dVx1 dx1is the radial flow field gradient; Vx1 is the radar radial flow field; los is the radar viewing angle.

[0024] As a preferred solution, the sea surface simulation is performed based on each of the radar flow field parameters to generate several sea surface backscatter coefficient-Doppler velocity combinations, including:

[0025] Obtain the radar viewing angle and radar band corresponding to each radar flow field parameter respectively;

[0026] For each radar flow field parameter, the radar radial flow field and the radial flow field gradient are input into a preset M4S model, so that the preset M4S model obtains the sea surface backscatter coefficient-Doppler velocity combination through the following steps:

[0027] Calculate sea surface roughness distribution based on radar radial flow field and radial flow field gradient;

[0028] Calculating backscatter coefficients of the sea surface in a plurality of different directions according to the sea surface roughness distribution and the radar viewing angle;

[0029] The backscatter coefficient of the sea surface is obtained by processing several backscatter coefficients using the integration method or the averaging method;

[0030] Based on the radar radial flow field, the radial flow field gradient and the radar band, the frequency shift of the sea surface echo signal is calculated using the Doppler effect formula;

[0031] Based on the radar viewing angle, converting the frequency shift of the sea surface echo signal into Doppler velocity;

[0032] The sea surface backscatter coefficient and Doppler velocity are combined to form a sea surface backscatter coefficient-Doppler velocity combination.

[0033] As a preferred solution, the method of performing regression prediction training on a fully connected neural network using each of the data pairs as training samples includes:

[0034] Using each of the data pairs as training samples, the fully connected neural network is controlled to cycle through the following steps:

[0035] The hidden layer of the fully connected neural network is trained to calculate the latent features based on the wind speed, radar polarization mode, radar band, radar incident angle, radar viewing angle, radar radial flow field and radial flow field gradient in the data pair;

[0036] The output layer of the fully connected neural network is trained to obtain a predicted sea surface backscatter coefficient-Doppler velocity based on the latent features;

[0037] Calculating a difference between a sea surface backscatter coefficient-Doppler velocity combination in the data pair and the predicted sea surface backscatter coefficient-Doppler velocity;

[0038] The weight value of each neuron in the fully connected neural network is updated based on the difference value.

[0039] As a preferred solution, the updating of the weight value of each neuron in the fully connected neural network based on the difference value includes:

[0040] The Levenberg-Marquardt algorithm is used to optimize the loss function of the fully connected neural network and determine the minimum value of the loss function;

[0041] The weight value of each neuron in the fully connected neural network is updated based on the minimum value of the loss function.

[0042] As a preferred solution, the fully connected neural network trained for regression prediction is determined as a sea surface prediction model, including:

[0043] When the mean square error between the sea surface backscatter coefficient-Doppler velocity combination in the data pair and the predicted sea surface backscatter coefficient-Doppler velocity is less than a preset condition threshold or the number of model cycle training reaches a preset round threshold, it is determined that the fully connected neural network completes the regression prediction training, and the fully connected neural network at this time is determined as the sea surface prediction model.

[0044] Accordingly, the present invention provides a device for predicting sea surface backscatter coefficient and Doppler velocity, comprising: a parameter setting module, a flow field simulation module, a parameter conversion module, a sea surface simulation module, a data combination module, a model training module and a prediction module;

[0045] The parameter setting module is used to randomly set several groups of sea surface data; wherein the sea surface data includes wind speed, radar polarization mode, radar band, radar incident angle and radar viewing angle;

[0046] The flow field simulation module is used to perform flow field simulation based on each of the sea surface data to obtain a plurality of groups of latitude and longitude flow fields; wherein each group of latitude and longitude flow fields includes a longitude flow field and a latitude flow field;

[0047] The parameter conversion module is used to convert each of the latitude and longitude flow fields into a plurality of groups of radar flow field parameters; wherein each group of radar flow field parameters includes a radar radial flow field and a radial flow field gradient;

[0048] The sea surface simulation module is used to perform sea surface simulation based on each of the radar flow field parameters to generate a plurality of sea surface backscatter coefficient-Doppler velocity combinations;

[0049] The data combination module is used to combine several groups of sea surface data, several groups of radar flow field parameters and several sea surface backscatter coefficient-Doppler velocity combinations according to corresponding relationships to form several data pairs;

[0050] The model training module is used to perform regression prediction training on the fully connected neural network using each of the data pairs as training samples, and determine the fully connected neural network after the regression prediction training as the sea surface prediction model;

[0051] The prediction module is used to input real-time wind speed, real-time radar polarization mode, real-time radar band, real-time radar incident angle, real-time radar viewing angle, real-time radar radial flow field and real-time radial flow field gradient into the sea surface prediction model to obtain a sea surface backscatter coefficient prediction value and a Doppler velocity prediction value.

[0052] The present invention also provides a terminal device, comprising: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps of the sea surface backscatter coefficient and Doppler velocity prediction method of the present invention are implemented.

[0053] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0054] The present invention provides a method for predicting sea surface backscatter coefficient and Doppler velocity. The method comprises the following steps: randomly setting a plurality of groups of sea surface data; performing flow field simulation based on the sea surface data to obtain a longitude flow field and a latitude flow field; converting the longitude flow field and the latitude flow field into a radar radial flow field and a radial flow field gradient; performing sea surface simulation based on the radar radial flow field and the radial flow field gradient to generate a sea surface backscatter coefficient-Doppler velocity combination; combining the sea surface data, the radar radial flow field and the radial flow field gradient, and the sea surface backscatter coefficient-Doppler velocity combination to form data pairs serving as training samples for a sea surface prediction model; and training a sea surface prediction model based on these data pairs to establish mapping relationships between multiple factors such as sea surface flow field, wind field, radar information, and the sea surface backscatter coefficient and Doppler velocity. The sea surface prediction model can be used to quickly predict the sea surface backscatter coefficient and Doppler velocity, thereby better adapting to rapid changes in the ocean environment, improving the efficiency of extracting key parameters from a complex ocean environment, and solving the problem of low efficiency in processing ocean remote sensing parameters in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solution of the present application, the following is a brief introduction to the drawings required for use in the implementation. Obviously, the drawings described below are only some implementation methods of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0056] Figure 1 A schematic flow chart of an embodiment of a method for predicting sea surface backscatter coefficient and Doppler velocity provided by the present invention;

[0057] Figure 2 A schematic diagram of a flow chart of the sea surface prediction model training method provided by the present invention;

[0058] Figure 3 A schematic structural diagram of an embodiment of a device for predicting sea surface backscatter coefficient and Doppler velocity provided by the present invention;

[0059] Figure 4 This is a structural schematic diagram of an embodiment of a device for predicting sea surface backscatter coefficient and Doppler velocity provided by the present invention. DETAILED DESCRIPTION

[0060] To make the objectives, technical solutions, and advantages of this application more clear, the technical solutions in this application will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative efforts are within the scope of protection of this application.

[0061] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0062] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0063] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0064] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0065] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0066] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0067] See also Figure 1 To solve the problem of low efficiency in processing ocean remote sensing parameters in the prior art, an embodiment of the present invention provides a method for predicting sea surface backscatter coefficient and Doppler velocity. The method includes steps 101 to 107, each of which is specifically as follows:

[0068] Step 101: randomly setting several groups of sea surface data; wherein the sea surface data includes wind speed, radar polarization mode, radar band, radar incident angle and radar viewing angle.

[0069] In an embodiment of the present invention, a sea surface prediction model is formed by training a fully connected neural network. The sea surface prediction model can be used to quickly obtain the sea surface backscatter coefficient and Doppler velocity by comprehensively considering multiple factors such as the sea surface flow field, wind field, and radar information. Training the fully connected neural network first requires collecting a certain number of training samples. Each training sample must include wind speed, radar polarization mode, radar band, radar incident angle, radar viewing angle, radar flow field parameters, as well as sea surface backscatter coefficient and Doppler velocity. Sea surface data such as wind speed, radar polarization mode, radar band, radar incident angle, and radar viewing angle can be randomly generated within a pre-set value range.

[0070] As an example of an embodiment of the present invention, the wind speed may range from 3 m / s to 12 m / s; the radar polarization modes may include horizontal transmission and horizontal reception (HH) and vertical transmission and vertical reception (VV); the radar bands may include L-band, C-band, X-band, Ka-band, and Ku-band; the radar incident angle may range from 20° to 40°; and the radar viewing angle may range from 0° to 360°.

[0071] Step 102: Flow field simulation is performed based on each of the sea surface data to obtain several groups of latitude and longitude flow fields; wherein each group of latitude and longitude flow fields includes a longitude flow field and a latitude flow field.

[0072] As a preferred solution of this embodiment, flow field simulation is performed based on each of the sea surface data to obtain several sets of latitude and longitude flow fields, including:

[0073] For each sea surface data, the sea surface data is input into the Princeton Ocean Model, so that the Princeton Ocean Model performs simulation according to the wind speed, radar polarization mode, radar band, radar incidence angle and radar viewing angle in the sea surface data to obtain flow velocity data at several different sea surface positions, and then analyzes the flow velocity data at several different sea surface positions to obtain longitude flow field and latitude flow field, and then generates latitude and longitude flow field according to the longitude flow field and the latitude flow field.

[0074] In an embodiment of the present invention, the Princeton Ocean Model (POM) is a three-dimensional baroclinic primitive equation numerical ocean model that can simulate various phenomena and processes in the ocean, including ocean circulation, tidal currents, wind-driven currents, etc. When simulating ocean circulation, the model can calculate the water flow speed and direction at different positions and depths, thereby obtaining the flow field information in the longitude and latitude directions. Therefore, the randomly generated sea surface data are input into the Princeton Ocean Model. The Princeton Ocean Model is based on the three-dimensional baroclinic primitive equation. By solving these equations, the velocity field in the ocean is obtained, including the velocity data in the longitude (east-west direction) and the latitude (north-south direction). By analyzing the velocity components in the longitude and latitude directions at each sea surface position, the longitude flow field and the latitude flow field can be obtained, thereby forming the longitude and latitude flow field. Wherein, the value range of the longitude flow field and the latitude flow field can be set in the Princeton Ocean Model. For example, the value range of the longitude flow field is set to 0.1m / sˉ1.2m / s, and the value range of the latitude flow field is set to 0.1m / sˉ1.2m / s.

[0075] Step 103: Convert each of the latitude and longitude flow fields into a plurality of groups of radar flow field parameters respectively; wherein each group of radar flow field parameters includes a radar radial flow field and a radial flow field gradient.

[0076] In an embodiment of the present invention, the characteristics of the radial flow field affect the scattering properties of electromagnetic waves on the sea surface. Different radial flow field velocities and gradients can cause changes in the intensity and frequency of scattered echoes. By focusing on the radial flow field, the relationship between these changes and the sea surface backscatter coefficient can be studied in more detail. With respect to Doppler velocity, radar primarily detects the Doppler shift caused by the target's radial velocity. Therefore, using the radar radial flow field and gradient can more directly calculate velocity information related to the Doppler shift, improving the accuracy and effectiveness of the calculation. Therefore, converting the longitude and latitude flow field into radar flow field parameters, including the radar radial flow field and radial flow field gradient, can more conveniently be integrated and analyzed collaboratively with other data based on radar observations or radial coordinates, thereby increasing the comprehensive utilization value of the data and providing more powerful support for applications such as marine environmental monitoring and marine resource development.

[0077] As a preferred solution of this embodiment, each of the latitude and longitude flow fields is converted into several sets of radar flow field parameters, including:

[0078] Obtain the radar viewing angle corresponding to each longitude and latitude flow field respectively;

[0079] For each latitude and longitude flow field, the radar radial flow field is calculated using the following formula:

[0080] Vx1=cos(los)*Vx0+sin(los)*Vy0

[0081] Where Vx1 is the radar radial flow field; Vx0 is the latitude flow field; Vy0 is the longitude flow field; and los is the radar viewing angle.

[0082] In an embodiment of the present invention, the corresponding radar viewing angle is determined for each longitude and latitude flow field. According to the above formula, the radar radial flow field can be calculated by combining the longitude flow field, the latitude flow field and the radar viewing angle.

[0083] As a preferred solution of this embodiment, each of the latitude and longitude flow fields is converted into several sets of radar flow field parameters, including:

[0084] For each latitude and longitude flow field, the radial flow field gradient is calculated using the following formula:

[0085]

[0086] Where, dVx1 dx1 is the radial flow field gradient; Vx1 is the radar radial flow field; los is the radar viewing angle.

[0087] In the embodiment of the present invention, after the radar radial flow field is calculated, the radial flow field gradient can be calculated according to the above formula in combination with the corresponding radar viewing angle.

[0088] Step 104: performing sea surface simulation based on each of the radar flow field parameters to generate a plurality of sea surface backscatter coefficient-Doppler velocity combinations.

[0089] As a preferred solution of this embodiment, sea surface simulation is performed based on each of the radar flow field parameters to generate several sea surface backscatter coefficient-Doppler velocity combinations, including:

[0090] Obtain the radar viewing angle and radar band corresponding to each radar flow field parameter respectively;

[0091] For each radar flow field parameter, the radar radial flow field and the radial flow field gradient are input into a preset M4S model, so that the preset M4S model obtains the sea surface backscatter coefficient-Doppler velocity combination through the following steps:

[0092] Calculate sea surface roughness distribution based on radar radial flow field and radial flow field gradient;

[0093] Calculating backscatter coefficients of the sea surface in a plurality of different directions according to the sea surface roughness distribution and the radar viewing angle;

[0094] The backscatter coefficient of the sea surface is obtained by processing several backscatter coefficients using the integration method or the averaging method;

[0095] Based on the radar radial flow field, the radial flow field gradient and the radar band, the frequency shift of the sea surface echo signal is calculated using the Doppler effect formula;

[0096] Based on the radar viewing angle, converting the frequency shift of the sea surface echo signal into Doppler velocity;

[0097] The sea surface backscatter coefficient and Doppler velocity are combined to form a sea surface backscatter coefficient-Doppler velocity combination.

[0098] In an embodiment of the present invention, based on randomly set sea surface data and radar radial flow field and radial flow field gradient obtained by simulating the sea surface data, the M4S model is utilized to perform sea surface simulation, and a sea surface backscatter coefficient-Doppler velocity combination can be obtained. Specifically, the model parameters of the preset M4S model are loaded, including sea surface roughness parameters, dielectric constant model, radar system parameters (such as transmission frequency, polarization mode, etc.) and other relevant physical constants and empirical parameters. The simulated sea surface area range is determined and divided into suitable grids. The resolution of the grid should be determined according to the accuracy of the input data and the requirements of the model calculation, so as to ensure that the flow field changes can be accurately captured while avoiding excessive calculation. Based on the input radar radial flow field and radial flow field gradient data, the initial state of the sea surface is estimated to obtain the sea surface roughness distribution.

[0099] In an embodiment of the present invention, after obtaining the sea surface roughness distribution, the backscatter coefficients of the sea surface in different directions are calculated by combining radar system parameters and electromagnetic scattering theory (such as Kirchhoff's approximation and perturbation theory). The M4S model typically considers multiple scattering mechanisms, including specular reflection and Bragg scattering. Therefore, after obtaining the backscatter coefficients of the sea surface in different directions, the contributions of various scattering mechanisms can be weighted summed or averaged according to the sea surface roughness and radar observation angle to obtain the sea surface backscatter coefficients.

[0100] In an embodiment of the present invention, the M4S model obtains the radar band corresponding to the radar radial flow field and radial flow field gradient. Based on the radar band, the radar transmit frequency can be determined. Combining the radar radial flow field, radial flow field gradient, and radar transmit frequency, the frequency shift of the sea surface echo signal can be calculated using the Doppler effect formula. The Doppler shift is proportional to the radial velocity of the sea surface, and the frequency shift can be calculated using known radar parameters and flow field velocity.

[0101] In an embodiment of the present invention, after calculating the frequency shift of the sea surface echo signal, the frequency shift is converted into Doppler velocity based on the radar's line of sight angle corresponding to the radar's radial flow field and radial flow field gradient. For example, by substituting the frequency shift of the sea surface echo signal into the Doppler velocity calculation formula and combining relevant geometric and kinematic relationships, the Doppler velocity of the sea surface in the radar's line of sight can be calculated.

[0102] In an embodiment of the present invention, after the sea surface backscatter coefficient and Doppler velocity are respectively obtained, the sea surface backscatter coefficient and Doppler velocity calculated based on the same set of radar flow field parameters are combined to form a sea surface backscatter coefficient-Doppler velocity combination.

[0103] Step 105: combining several groups of sea surface data, several groups of radar flow field parameters and several sea surface backscatter coefficient-Doppler velocity combinations according to corresponding relationships to form several data pairs.

[0104] In an embodiment of the present invention, the longitude flow field and the latitude flow field are simulated based on sea surface data. Converting the longitude flow field and the latitude flow field can generate a radar radial flow field and a radial flow field gradient. Simulating the radar radial flow field and the radial flow field gradient on the sea surface can generate a sea surface backscatter coefficient-Doppler velocity combination. Therefore, a set of sea surface data corresponds to a set of radar radial flow fields and radial flow field gradients, and a set of sea surface backscatter coefficient-Doppler velocity combinations. Based on the corresponding relationship between the data, the corresponding sea surface data, radar flow field parameters, and sea surface backscatter coefficient-Doppler velocity combinations are combined to generate multiple data pairs. The resulting data pairs can be used as training samples for training a sea surface prediction model.

[0105] Step 106: Using each of the data pairs as training samples, perform regression prediction training on the fully connected neural network, and determine the fully connected neural network after the regression prediction training as the sea surface prediction model.

[0106] As a preferred solution of this embodiment, each of the data pairs is used as a training sample to perform regression prediction training on a fully connected neural network, including:

[0107] Using each of the data pairs as training samples, the fully connected neural network is controlled to cycle through the following steps:

[0108] The hidden layer of the fully connected neural network is trained to calculate the latent features based on the wind speed, radar polarization mode, radar band, radar incident angle, radar viewing angle, radar radial flow field and radial flow field gradient in the data pair;

[0109] The output layer of the fully connected neural network is trained to obtain a predicted sea surface backscatter coefficient-Doppler velocity based on the latent features;

[0110] Calculating a difference between a sea surface backscatter coefficient-Doppler velocity combination in the data pair and the predicted sea surface backscatter coefficient-Doppler velocity;

[0111] The weight value of each neuron in the fully connected neural network is updated based on the difference value.

[0112] In an embodiment of the present invention, the obtained data pairs are used as training samples to train a fully connected neural network (DNN) for regression prediction. The structure of the deep neural network (DNN) includes an input layer, three hidden layers, and an output layer. The activation function of the three hidden layers is Tanh. Tanh is a nonlinear function with an output range of [-1, 1]. It has good symmetry and gradient characteristics, which helps improve the convergence speed and performance of the neural network. The specific training process is as follows: During the feedforward propagation process, the hidden layer of the deep neural network calculates latent features based on wind speed, radar polarization mode, radar band, radar incidence angle, radar viewing angle, radar radial flow field, and radial flow field gradient. The output layer uses the latent features to derive a regression prediction value, namely, the predicted sea surface backscatter coefficient (Doppler velocity). During the backpropagation process, the difference between the actual value and the predicted value is calculated, and a related algorithm is used to update the weight value of each neuron in the fully connected neural network based on this difference. This process is repeated continuously to train the fully connected neural network.

[0113] As a preferred solution of this embodiment, updating the weight value of each neuron in the fully connected neural network based on the difference value includes:

[0114] The Levenberg-Marquardt algorithm is used to optimize the loss function of the fully connected neural network and determine the minimum value of the loss function;

[0115] The weight value of each neuron in the fully connected neural network is updated based on the minimum value of the loss function.

[0116] In an embodiment of the present invention, the Levenberg-Marquardt algorithm is an optimization algorithm for nonlinear least squares problems. The algorithm is used to adjust the weights of neurons in the model. It can combine the advantages of the gradient descent method and the Newton method, efficiently find the minimum value of the loss function, and improve the model convergence speed.

[0117] As a preferred solution of this embodiment, the fully connected neural network after regression prediction training is determined as the sea surface prediction model, including:

[0118] When the mean square error between the sea surface backscatter coefficient-Doppler velocity combination in the data pair and the predicted sea surface backscatter coefficient-Doppler velocity is less than a preset condition threshold or the number of model cycle training reaches a preset round threshold, it is determined that the fully connected neural network completes the regression prediction training, and the fully connected neural network at this time is determined as the sea surface prediction model.

[0119] In an embodiment of the present invention, the mean square error (MSE) is used as a loss function to ensure that the model can effectively reduce the prediction error during the training process. Therefore, during each model training, the mean square error (MSE) between the true value and the predicted value is calculated. When the mean square error (MSE) is less than a preset condition threshold, or the number of cycle training of the model reaches a preset round threshold, the model training is determined to be complete. As an example, the preset condition threshold can be set to 0, and the preset round threshold can be set to 1000 times.

[0120] See also Figure 2 , is a flow chart of the sea surface prediction model training method provided by the present invention. Based on a given data set D, i.e., multiple data pairs generated, data processing is performed to form an input matrix X that conforms to the model input data format; the input matrix X is input to a fully connected neural network, the network is initialized, and training data is allocated according to the input matrix X; the fully connected neural network is trained based on the training data, and the weight parameter matrix W is updated. i , and at the same time calculate the mean square error (MSE) between the true value and the predicted value. When the mean square error (MSE) reaches 0, or the number of cyclic training reaches 1000, the training of the fully connected neural network is determined to be completed, and the sea surface prediction model is formed by verifying the network performance of the fully connected neural network.

[0121] Step 107: Input the real-time wind speed, real-time radar polarization mode, real-time radar band, real-time radar incident angle, real-time radar viewing angle, real-time radar radial flow field, and real-time radial flow field gradient into the sea surface prediction model to obtain a predicted value of the sea surface backscatter coefficient and a predicted value of the Doppler velocity.

[0122] In an embodiment of the present invention, the sea surface prediction model establishes a mapping relationship between various factors, such as the sea surface flow field, wind field, and radar information, and the sea surface backscatter coefficient and Doppler velocity. Therefore, when acquiring real-time wind speed, radar polarization mode, radar band, radar incidence angle, radar viewing angle, radar radial flow field, and radial flow field gradient, and inputting this data into the sea surface prediction model, the sea surface backscatter coefficient and Doppler velocity can be rapidly derived, improving the efficiency of extracting key parameters from complex ocean environments.

[0123] The implementation of the above embodiment has the following effects:

[0124] The present invention provides a method for predicting sea surface backscatter coefficient and Doppler velocity. The method comprises the following steps: randomly setting a plurality of groups of sea surface data; performing flow field simulation based on the sea surface data to obtain a longitude flow field and a latitude flow field; converting the longitude flow field and the latitude flow field into a radar radial flow field and a radial flow field gradient; performing sea surface simulation based on the radar radial flow field and the radial flow field gradient to generate a sea surface backscatter coefficient-Doppler velocity combination; combining the sea surface data, the radar radial flow field and the radial flow field gradient, and the sea surface backscatter coefficient-Doppler velocity combination to form data pairs serving as training samples for a sea surface prediction model; and training a sea surface prediction model based on these data pairs to establish mapping relationships between multiple factors such as sea surface flow field, wind field, radar information, and the sea surface backscatter coefficient and Doppler velocity. The sea surface prediction model can be used to quickly predict the sea surface backscatter coefficient and Doppler velocity, thereby better adapting to rapid changes in the ocean environment, improving the efficiency of extracting key parameters from a complex ocean environment, and solving the problem of low efficiency in processing ocean remote sensing parameters in the prior art.

[0125] like Figure 3 As shown, based on the above method embodiment, a corresponding device embodiment is provided;

[0126] An embodiment of the present invention provides a device for predicting sea surface backscatter coefficient and Doppler velocity, comprising: a parameter setting module, a flow field simulation module, a parameter conversion module, a sea surface simulation module, a data combination module, a model training module, and a prediction module;

[0127] The parameter setting module is used to randomly set several groups of sea surface data; wherein the sea surface data includes wind speed, radar polarization mode, radar band, radar incident angle and radar viewing angle;

[0128] The flow field simulation module is used to perform flow field simulation based on each of the sea surface data to obtain a plurality of groups of latitude and longitude flow fields; wherein each group of latitude and longitude flow fields includes a longitude flow field and a latitude flow field;

[0129] The parameter conversion module is used to convert each of the latitude and longitude flow fields into a plurality of groups of radar flow field parameters; wherein each group of radar flow field parameters includes a radar radial flow field and a radial flow field gradient;

[0130] The sea surface simulation module is used to perform sea surface simulation based on each of the radar flow field parameters to generate a plurality of sea surface backscatter coefficient-Doppler velocity combinations;

[0131] The data combination module is used to combine several groups of sea surface data, several groups of radar flow field parameters and several sea surface backscatter coefficient-Doppler velocity combinations according to corresponding relationships to form several data pairs;

[0132] The model training module is used to perform regression prediction training on the fully connected neural network using each of the data pairs as training samples, and determine the fully connected neural network after the regression prediction training as the sea surface prediction model;

[0133] The prediction module is used to input real-time wind speed, real-time radar polarization mode, real-time radar band, real-time radar incident angle, real-time radar viewing angle, real-time radar radial flow field and real-time radial flow field gradient into the sea surface prediction model to obtain a sea surface backscatter coefficient prediction value and a Doppler velocity prediction value.

[0134] It can be understood that the above-mentioned device embodiment corresponds to the method embodiment of the present invention, and can implement any of the above-mentioned method embodiments of the present invention to provide a method for predicting sea surface backscatter coefficient and Doppler velocity.

[0135] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. Furthermore, in the drawings of the device embodiments provided by the present invention, the connection relationship between modules indicates that they have a communication connection, which may be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement the present invention without inventive effort.

[0136] Based on the above-mentioned embodiments of the sea surface backscatter coefficient and Doppler velocity prediction method, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements the sea surface backscatter coefficient and Doppler velocity prediction method of any embodiment of the present invention.

[0137] See also Figure 4 , is a structural diagram of an embodiment of the sea surface backscatter coefficient and Doppler velocity prediction device provided by the present invention. The device can be a server, including a processor, a memory and a network interface connected through a system bus, wherein the memory can include a non-volatile storage medium and an internal memory.

[0138] The processor is used to provide computing and control capabilities and support the operation of the entire computer equipment.

[0139] The non-volatile storage medium can store an operating system and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any one of the sea surface backscatter coefficient and Doppler velocity prediction methods.

[0140] The internal memory provides an environment for the operation of the computer program in the non-volatile storage medium. When the computer program is executed by the processor, the processor can execute any sea surface backscatter coefficient and Doppler velocity prediction method.

[0141] The network interface is used for network communication, such as sending assigned tasks.

[0142] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0143] It should be understood that the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0144] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that 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 for those skilled in the art.

Claims

1. A method for predicting sea surface backscatter coefficient and Doppler velocity, characterized in that: include: Randomly setting a number of groups of sea surface data; wherein the sea surface data includes wind speed, radar polarization mode, radar band, radar incident angle and radar viewing angle; Performing flow field simulation based on each of the sea surface data to obtain several groups of latitude and longitude flow fields; wherein each group of latitude and longitude flow fields includes a longitude flow field and a latitude flow field; Converting each of the latitude and longitude flow fields into a plurality of groups of radar flow field parameters; wherein each group of radar flow field parameters includes a radar radial flow field and a radial flow field gradient; Performing sea surface simulation based on each of the radar flow field parameters to generate a plurality of sea surface backscatter coefficient-Doppler velocity combinations; Combining several sets of sea surface data, several sets of radar flow field parameters and several sea surface backscatter coefficient-Doppler velocity combinations according to corresponding relationships to form several data pairs; Using each of the data pairs as training samples, performing regression prediction training on a fully connected neural network, and determining the fully connected neural network after the regression prediction training as a sea surface prediction model; The real-time wind speed, real-time radar polarization mode, real-time radar band, real-time radar incident angle, real-time radar viewing angle, real-time radar radial flow field and real-time radial flow field gradient are input into the sea surface prediction model to obtain a sea surface backscatter coefficient prediction value and a Doppler velocity prediction value.

2. The method for predicting sea surface backscatter coefficient and Doppler velocity according to claim 1, wherein: The flow field simulation is performed based on each of the sea surface data to obtain several sets of latitude and longitude flow fields, including: For each sea surface data, the sea surface data is input into the Princeton Ocean Model, so that the Princeton Ocean Model performs simulation according to the wind speed, radar polarization mode, radar band, radar incidence angle and radar viewing angle in the sea surface data to obtain flow velocity data at several different sea surface positions, and then analyzes the flow velocity data at several different sea surface positions to obtain longitude flow field and latitude flow field, and then generates latitude and longitude flow field according to the longitude flow field and the latitude flow field.

3. The method for predicting sea surface backscatter coefficient and Doppler velocity according to claim 1, wherein: The step of converting the latitude and longitude flow fields into a plurality of sets of radar flow field parameters comprises: Obtain the radar viewing angle corresponding to each longitude and latitude flow field respectively; For each latitude and longitude flow field, the radar radial flow field is calculated using the following formula: Vx1=cos(los)*Vx0+sin(los)*Vy0 Where Vx1 is the radar radial flow field; Vx0 is the latitude flow field; Vy0 is the longitude flow field; and los is the radar viewing angle.

4. The method for predicting sea surface backscatter coefficient and Doppler velocity according to claim 3, wherein: The step of converting the latitude and longitude flow fields into a plurality of sets of radar flow field parameters comprises: For each latitude and longitude flow field, the radial flow field gradient is calculated using the following formula: Where, dVx1 dx1 is the radial flow field gradient; Vx1 is the radar radial flow field; los is the radar viewing angle.

5. The method for predicting sea surface backscatter coefficient and Doppler velocity according to claim 1, wherein: The sea surface simulation is performed based on each of the radar flow field parameters to generate a plurality of sea surface backscatter coefficient-Doppler velocity combinations, including: Obtain the radar viewing angle and radar band corresponding to each radar flow field parameter respectively; For each radar flow field parameter, the radar radial flow field and the radial flow field gradient are input into a preset M4S model, so that the preset M4S model obtains the sea surface backscatter coefficient-Doppler velocity combination through the following steps: Calculate sea surface roughness distribution based on radar radial flow field and radial flow field gradient; Calculating backscatter coefficients of the sea surface in a plurality of different directions according to the sea surface roughness distribution and the radar viewing angle; The backscatter coefficient of the sea surface is obtained by processing several backscatter coefficients using the integration method or the averaging method; Based on the radar radial flow field, the radial flow field gradient and the radar band, the frequency shift of the sea surface echo signal is calculated using the Doppler effect formula; Based on the radar viewing angle, converting the frequency shift of the sea surface echo signal into Doppler velocity; The sea surface backscatter coefficient and Doppler velocity are combined to form a sea surface backscatter coefficient-Doppler velocity combination.

6. The method for predicting sea surface backscatter coefficient and Doppler velocity according to claim 1, wherein: The method of performing regression prediction training on a fully connected neural network using each of the data pairs as training samples includes: Using each of the data pairs as training samples, the fully connected neural network is controlled to cycle through the following steps: The hidden layer of the fully connected neural network is trained to calculate the latent features based on the wind speed, radar polarization mode, radar band, radar incident angle, radar viewing angle, radar radial flow field and radial flow field gradient in the data pair; The output layer of the fully connected neural network is trained to obtain a predicted sea surface backscatter coefficient-Doppler velocity based on the latent features; Calculating a difference between a sea surface backscatter coefficient-Doppler velocity combination in the data pair and the predicted sea surface backscatter coefficient-Doppler velocity; The weight value of each neuron in the fully connected neural network is updated based on the difference value.

7. The method for predicting sea surface backscatter coefficient and Doppler velocity according to claim 6, wherein: The updating of the weight value of each neuron in the fully connected neural network based on the difference value includes: The Levenberg-Marquardt algorithm is used to optimize the loss function of the fully connected neural network and determine the minimum value of the loss function; The weight value of each neuron in the fully connected neural network is updated based on the minimum value of the loss function.

8. The method for predicting sea surface backscatter coefficient and Doppler velocity according to claim 6, wherein: The step of determining the fully connected neural network that has completed regression prediction training as the sea surface prediction model includes: When the mean square error between the sea surface backscatter coefficient-Doppler velocity combination in the data pair and the predicted sea surface backscatter coefficient-Doppler velocity is less than a preset condition threshold or the number of model cycle training reaches a preset round threshold, it is determined that the fully connected neural network completes the regression prediction training, and the fully connected neural network at this time is determined as the sea surface prediction model.

9. A device for predicting sea surface backscatter coefficient and Doppler velocity, characterized in that: include: Parameter setting module, flow field simulation module, parameter conversion module, sea surface simulation module, data combination module, model training module and prediction module; The parameter setting module is used to randomly set several groups of sea surface data; wherein the sea surface data includes wind speed, radar polarization mode, radar band, radar incident angle and radar viewing angle; The flow field simulation module is used to perform flow field simulation based on each of the sea surface data to obtain a plurality of groups of latitude and longitude flow fields; wherein each group of latitude and longitude flow fields includes a longitude flow field and a latitude flow field; The parameter conversion module is used to convert each of the latitude and longitude flow fields into a plurality of groups of radar flow field parameters; wherein each group of radar flow field parameters includes a radar radial flow field and a radial flow field gradient; The sea surface simulation module is used to perform sea surface simulation based on each of the radar flow field parameters to generate a plurality of sea surface backscatter coefficient-Doppler velocity combinations; The data combination module is used to combine several groups of sea surface data, several groups of radar flow field parameters and several sea surface backscatter coefficient-Doppler velocity combinations according to corresponding relationships to form several data pairs; The model training module is used to perform regression prediction training on the fully connected neural network using each of the data pairs as training samples, and determine the fully connected neural network after the regression prediction training as the sea surface prediction model; The prediction module is used to input real-time wind speed, real-time radar polarization mode, real-time radar band, real-time radar incident angle, real-time radar viewing angle, real-time radar radial flow field and real-time radial flow field gradient into the sea surface prediction model to obtain a sea surface backscatter coefficient prediction value and a Doppler velocity prediction value.

10. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for predicting the sea surface backscatter coefficient and Doppler velocity according to any one of claims 1 to 8 is implemented.

Images