A method, apparatus, medium, and computing equipment for joint inversion of sea surface wind speed

Abstract

This invention discloses a joint inversion method for sea surface wind speed, comprising: jointly inverting observation data from a microwave scatterometer and a spaceborne GNSS-R over an overlapping observation area to obtain a first wind speed inversion result for that area; independently inverting the observation data from the microwave scatterometer to obtain a second wind speed inversion result; expanding the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer; independently inverting the observation data from a synthetic aperture radar to obtain a fourth wind speed inversion result; and expanding the fourth wind speed inversion result to the third wind speed inversion result to obtain a final joint inversion result. The technical solution provided by this invention performs joint inversion based on observation data from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar, thereby obtaining a final joint inversion result with a large observation area, high accuracy, and high resolution.

Description

Technical Field

[0001] This invention relates to the field of sea surface wind speed observation, and in particular to a method, apparatus, medium, and computing equipment for joint inversion of sea surface wind speed. Background Technology

[0002] Currently, global sea surface wind speed joint inversion products have wide applications in meteorology and oceanography. Typhoons are among the most destructive natural disasters, and real-time observations of typhoon paths and intensities are required through satellite remote sensing in disaster early warning and rescue applications. Currently, satellite remote sensing can be used to observe sea surface wind speeds using various methods, such as microwave scatterometers, spaceborne GNSS-R, and SAR sensors. Microwave scatterometers offer a large observation range, enabling global observation and penetrating clouds and fog, but are limited to low-wind-speed observations, with lower accuracy for high-wind-speed observations. Spaceborne GNSS-R offers high accuracy for high-wind-speed observations, but its observation range is limited, making global observation impossible with a single spaceborne GNSS-R. SAR sensors can provide all-weather observations and have high resolution, but their observation range is also limited, preventing global observation. Summary of the Invention

[0003] The main objective of this invention is to propose a method, apparatus, medium, and computing device for joint inversion of sea surface wind speed, aiming to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, this invention proposes a joint inversion method for sea surface wind speed, comprising multiple observation data obtained from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar, respectively. The observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer, and the microwave scatterometer and the spaceborne GNSS-R have at least partial overlap in their observation areas. The accuracy of the observation data from the spaceborne GNSS-R is higher than that from the microwave scatterometer, and the resolution of the observation data from the synthetic aperture radar is higher than that from the microwave scatterometer. The method includes:

[0005] Based on the observation data of the microwave scatterometer and the spaceborne GNSS-R within the overlapping observation area, a first wind speed inversion result is obtained by joint inversion within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the observation data of the spaceborne GNSS-R.

[0006] Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer.

[0007] The first wind speed inversion result is extended to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result;

[0008] Based on the observation data of the synthetic aperture radar, a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar is obtained by separate inversion, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar.

[0009] The fourth wind speed inversion result is extended to the third wind speed inversion result to obtain the final joint inversion result. The accuracy and observation area of ​​the final joint inversion result are consistent with those of the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with that of the fourth wind speed inversion result.

[0010] In this embodiment of the application, based on the observation data of the microwave scatterometer and the spaceborne GNSS-R overlapping observation area, a first wind speed inversion result is obtained by joint inversion within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, including:

[0011] Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, the first observation data of the microwave scatterometer and the second observation data of the spaceborne GNSS-R are acquired.

[0012] Based on the geophysical model of the microwave scatterometer, first simulated observation data corresponding to the first observation data is obtained;

[0013] Based on the geophysical model of the spaceborne GNSS-R, second simulated observation data corresponding to the second observation data is obtained;

[0014] The first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model.

[0015] In this embodiment of the application, the observation data of the spaceborne GNSS-R includes at least two of the following: mean time-delay Doppler correlation power, slope of the leading edge of the time-delay correlation curve, and slope of the trailing edge of the time-delay correlation curve.

[0016] In this embodiment of the application, obtaining the first wind speed inversion result based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model includes:

[0017] Based on the joint inversion model, the first error ratio of the observation data of the microwave scatterometer and the second error ratio of the observation data of the spaceborne GNSS-R are obtained respectively.

[0018] Based on the first error ratio and the second error ratio, as well as the preset weights of the microwave scatterometer observation data and the preset weights of the spaceborne GNSS-R observation data, the wind speed inversion results in the overlapping observation area of ​​the microwave scatterometer and the GNSS-R are obtained.

[0019] According to the preset wind speed search step size, the wind speed inversion result is iterated until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is taken as the first wind speed inversion result.

[0020] In this embodiment of the application, the step of expanding the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer includes:

[0021] The first wind speed inversion result is filled in based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result;

[0022] Based on the second wind speed inversion result and the filled and fused wind speed inversion result, as well as the preset first adversarial generation model, the third wind speed inversion result is obtained.

[0023] In this embodiment of the application, the first adversarial generative model is trained through the following steps:

[0024] Obtain wind speed results samples obtained from the inversion of observation data based solely on microwave scatterometers, as well as wind speed results samples after filling and fusion;

[0025] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model;

[0026] Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively;

[0027] The first adversarial generation model is optimized based on the adversarial loss, the reconstruction loss, and the allocation loss.

[0028] In this embodiment of the application, the step of expanding the fourth wind speed inversion result to the third wind speed inversion result to obtain the final joint inversion result includes:

[0029] Based on the preset second adversarial generation model, the final joint inversion result is generated using the fourth wind speed inversion result and the third wind speed inversion result.

[0030] In this embodiment of the application, the second adversarial generative model is trained through the following steps:

[0031] Obtain the third wind speed inversion result sample, as well as the wind speed result sample obtained solely from the observation data inverted from synthetic aperture radar;

[0032] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the third wind speed inversion result sample, the second fusion result is generated using the second adversarial generative model;

[0033] Calculate the adversarial loss and feature loss of the second fusion result;

[0034] The second adversarial generation model is optimized based on the adversarial loss and the feature loss.

[0035] In this embodiment of the application, the method further includes:

[0036] After obtaining the third wind speed inversion result, determine the authenticity of the third wind speed inversion result; and / or

[0037] After obtaining the final joint inversion result, the authenticity of the final joint inversion result is determined.

[0038] In this embodiment of the application, the authenticity of the inversion result is determined by the following method:

[0039] Get any wind speed product;

[0040] Spatial matching is performed on the inversion results to obtain the RMSE accuracy of the inversion results;

[0041] The authenticity of the wind speed inversion results is determined based on the RMSE accuracy of the inversion results.

[0042] This invention also proposes a joint inversion device for sea surface wind speed, based on multiple observation data from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar. The observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer, and the microwave scatterometer and the spaceborne GNSS-R have at least partially overlapping observation areas. The accuracy of the observation data from the spaceborne GNSS-R is higher than that of the observation data from the microwave scatterometer, and the resolution of the observation data from the synthetic aperture radar is higher than that of the observation data from the microwave scatterometer. The device includes:

[0043] The acquisition module is used to acquire the observation data of the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas, the observation data of the microwave scatterometer, and the observation data of the synthetic aperture radar.

[0044] The inversion module is used to jointly invert observation data from the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas to obtain a first wind speed inversion result for that area, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the spaceborne GNSS-R observation data; and

[0045] Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer.

[0046] The first adversarial generation module is used to expand the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result.

[0047] The inversion module is further configured to, based on the observation data of the synthetic aperture radar, separately invert to obtain a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar.

[0048] The second adversarial generation module is used to expand the fourth wind speed inversion result to the third wind speed inversion result to obtain a final joint inversion result, wherein the accuracy and observation area of ​​the final joint inversion result are consistent with the accuracy and observation area of ​​the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with the resolution of the fourth wind speed inversion result.

[0049] In this embodiment of the application, the acquisition module is configured as follows:

[0050] Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, first observation data from the microwave scatterometer and second observation data from the spaceborne GNSS-R are acquired; based on the geophysical model of the microwave scatterometer, first simulated observation data corresponding to the first observation data are acquired; and

[0051] Based on the geophysical model of the spaceborne GNSS-R, second simulated observation data corresponding to the second observation data is obtained;

[0052] The inversion module is configured as follows:

[0053] The first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model.

[0054] In this embodiment of the application, the acquisition module is further configured to: when acquiring the observation data of the spaceborne GNSS-R, acquire at least two of the following: the mean value of time delay Doppler correlation power, the slope of the leading edge of the time delay correlation curve, and the slope of the trailing edge of the time delay correlation curve.

[0055] In this embodiment of the application, the inversion module is configured as follows:

[0056] Based on the joint inversion model, the first error ratio of the observation data of the microwave scatterometer and the second error ratio of the observation data of the spaceborne GNSS-R are obtained respectively.

[0057] Based on the first error ratio and the second error ratio, as well as the preset weights of the microwave scatterometer observation data and the preset weights of the spaceborne GNSS-R observation data, the wind speed inversion results in the overlapping observation area of ​​the microwave scatterometer and the GNSS-R are obtained.

[0058] According to the preset wind speed search step size, the wind speed inversion result is iterated until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is taken as the first wind speed inversion result.

[0059] In this embodiment of the application, the device further includes a filling module, which is configured to:

[0060] The first wind speed inversion result is filled in based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result;

[0061] The first adversarial generation module is configured as follows:

[0062] The filled and fused wind speed inversion result is expanded to the second wind speed inversion result to obtain the third wind speed inversion result.

[0063] In this embodiment of the application, the first adversarial generative model is trained as follows:

[0064] Obtain wind speed results samples obtained from the inversion of observation data based solely on microwave scatterometers, as well as wind speed results samples after filling and fusion;

[0065] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model;

[0066] Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively;

[0067] The first adversarial generation model is optimized based on the adversarial loss, the reconstruction loss, and the allocation loss.

[0068] In this embodiment of the application, the second adversary generation module is configured as follows:

[0069] The fourth wind speed inversion result is extended to the third wind speed inversion result to obtain the final joint inversion result.

[0070] In this embodiment of the application, the second adversarial generative model is trained as follows:

[0071] Obtain the third wind speed inversion result sample, as well as the wind speed result sample obtained solely from the observation data inverted from synthetic aperture radar;

[0072] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the third wind speed inversion result sample, the second fusion result is generated using the second adversarial generative model;

[0073] Calculate the adversarial loss and feature loss of the second fusion result;

[0074] The second adversarial generation model is optimized based on the adversarial loss and the feature loss.

[0075] In this embodiment of the application, the first adversarial generative model is further used to determine the authenticity of the third wind speed inversion result after obtaining the third wind speed inversion result;

[0076] The second adversarial generative model is also used to determine the authenticity of the final joint wind speed inversion result after obtaining the final joint wind speed inversion result.

[0077] In this embodiment of the application, the first adversarial generative model is further configured as follows:

[0078] Get any wind speed product;

[0079] Spatially match the wind speed product and the third wind speed inversion result to obtain the RMSE accuracy of the third wind speed inversion result;

[0080] Based on the RMSE accuracy of the third wind speed inversion result, the authenticity of the third wind speed inversion result is determined;

[0081] The second adversarial generative model is also configured as follows:

[0082] Get any wind speed product;

[0083] Spatially match the wind speed product and the final joint wind speed inversion result to obtain the RMSE accuracy of the final joint wind speed inversion result;

[0084] The authenticity of the final joint wind speed inversion result is determined based on the RMSE accuracy of the final joint wind speed inversion result.

[0085] The present invention also proposes a medium on which a computer program is stored, wherein the computer program, when executed by a processor, implements the method described in any of the preceding claims.

[0086] The present invention also proposes a computing device comprising a processor for executing a computer program stored in a memory to implement the method described in any of the preceding claims.

[0087] The technical solution provided by this invention first performs joint inversion based on observation data from the overlapping observation area of ​​a microwave scatterometer and a spaceborne GNSS-R, resulting in a first wind speed inversion result with a relatively small observation area and high accuracy. Then, the first wind speed inversion result is expanded onto a second wind speed inversion result obtained solely from observation data from the microwave scatterometer, resulting in a third wind speed inversion result with a larger observation area and high accuracy. Finally, a fourth wind speed inversion result obtained solely from observation data from a synthetic aperture radar is expanded onto the third wind speed inversion result, thereby obtaining a final joint inversion result with a larger observation area, higher accuracy, and higher resolution. Attached Figure Description

[0088] 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 the structures shown in these drawings without creative effort.

[0089] Figure 1 This is a schematic diagram of the observation area of ​​a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar in one embodiment of this application;

[0090] Figure 2 This is a flowchart illustrating the steps of the joint inversion method for sea surface wind speed in one embodiment of this application;

[0091] Figure 3 This is a schematic diagram of the structure of the first adversarial generative model in one embodiment of this application;

[0092] Figure 4 The image shows the third wind speed inversion result and the ERA5 wind speed result in one embodiment of this application.

[0093] Figure 5 for Figure 4 Scatter plot of the third wind speed inversion result and the ERA5 wind speed result after spatial matching;

[0094] Figure 6 This is a schematic diagram of the structure of the second adversarial generative model in one embodiment of this application;

[0095] Figure 7 This is a schematic diagram of the structure of the sea surface wind speed joint inversion device in one embodiment of this application;

[0096] Figure 8 This is a schematic diagram of the structure of the medium in one embodiment of this application;

[0097] Figure 9 This is a schematic diagram of the settlement device in one embodiment of this application.

[0098] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0099] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are given merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make this disclosure more thorough and complete, and to fully convey the scope of this disclosure to those skilled in the art.

[0100] Those skilled in the art will recognize that embodiments of the present invention can be implemented as an apparatus, device, method, or computer program product. Therefore, this disclosure can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0101] According to an embodiment of the present invention, a method, apparatus, medium and computing equipment for joint inversion of wind speed on sea surface are proposed. Invention Overview

[0103] The inventors discovered that microwave scatterometers are currently commonly used for large-scale global sea surface wind speed observations. The method of using microwave scatterometers to retrieve sea surface wind speed typically involves constructing a geophysical model function (GMF) using airborne or simulated data, and then using multiple observations of a specific point under actual observation conditions, employing methods such as maximum likelihood to ultimately determine the wind speed for that area. Microwave scatterometer retrieval of sea surface wind speed has the following advantages: 1) The frequency band used by microwave scatterometers, compared to the visible light band, can penetrate clouds and fog, allowing for normal data acquisition even in environments with cloud cover and at night, enabling all-weather observation. 2) Microwave scatterometers typically have a large scan swath, providing continuous observation coverage and the ability to achieve joint retrieval of global sea surface wind speeds within a single day.

[0104] Microwave scatterometers have a relatively long history, but the development of their algorithms and sensor technologies has been slow to date. Furthermore, because they rely on GMF (Gross Measuring Fluid) to retrieve sea surface wind speeds, their accuracy is poor in high-wind-speed scenarios, sometimes rendering them unusable. In applications such as typhoons, the retrieved wind speeds differ significantly from the true values.

[0105] Spaceborne GNSS-R is a low-cost method for joint inversion of sea surface wind speed, enabling high-precision observations in high-wind-speed scenarios. However, due to its random observation characteristics, it requires the launch of multiple small satellites to establish an observation starlink, aiming for maximum global coverage to achieve the goal of global observation.

[0106] Synthetic Aperture Radar (SAR) can provide high-resolution microwave imagery of the sea surface in real time, 24 / 7. Using the Normalized Scattering Cross Section (NRCS), sea surface wind speed can be observed and retrieved at kilometer resolution. High-resolution sea surface wind speed products are often used in coastal areas, such as for assessing coastal wind energy resources. However, due to the inherent characteristics of SAR sensors, their observation range is relatively small, and therefore, they generally cannot perform global sea surface wind speed observation and retrieval.

[0107] Generative Adversarial Networks (GANs) are machine learning architectures trained using adversarial strategies between a generative network (which randomly generates outputs) and a discriminative network (which identifies incorrect and correct samples). GANs have demonstrated superior performance in image and video generation, inpainting, and image super-resolution. Since sea surface wind speed inversion can be viewed as a series of snapshots of sea surface wind speeds, GANs can be used to fuse GNSS-R and SAR data on top of a microwave scatterometer, forming a new method for wind speed observation.

[0108] Therefore, this application proposes a joint inversion method using a microwave scatterometer, GNSS-R, and synthetic aperture radar, aiming to integrate the observation advantages of the three sensors and ultimately obtain high-precision, high-resolution joint inversion results of global sea surface wind speed.

[0109] After introducing the basic principles of the present invention, various non-limiting embodiments of the present invention will be described in detail below.

[0110] Exemplary methods

[0111] Please refer to the reference. Figure 1 This exemplary embodiment proposes a joint inversion method for sea surface wind speed. The method includes multiple observation data obtained from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar, respectively. The observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer. The microwave scatterometer and the spaceborne GNSS-R have at least partially overlapping observation areas, and the accuracy of the observation data from the spaceborne GNSS-R is higher than that of the observation data from the microwave scatterometer. The resolution of the observation data from the synthetic aperture radar is higher than that of the observation data from the microwave scatterometer.

[0112] like Figure 1 As shown, the microwave scatterometer has a large observation range. Figure 1 In the diagram, region S1 is the observation area of ​​the microwave scatterometer, S2 is the single-track observation area of ​​the spaceborne GNSS-R, and S3 is the observation area of ​​the synthetic aperture radar. Both S2 and S3 are smaller than S1, and both S2 and S3 are located within S1. Therefore, the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R is region S2, and the overlapping observation area of ​​the microwave scatterometer and the synthetic aperture radar is region S3. It should be noted that in other embodiments, S2 and S3 may only partially lie within S1.

[0113] like Figure 2 As shown in the embodiments of this application, the method includes the following steps:

[0114] Step S100: Based on the observation data of the microwave scatterometer and the spaceborne GNSS-R within the overlapping observation area, a first wind speed inversion result is obtained by joint inversion within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the observation data of the spaceborne GNSS-R.

[0115] Step S200: Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer.

[0116] Step S300: Expand the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result.

[0117] Step S400: Based on the observation data of the synthetic aperture radar, a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar is obtained by separate inversion, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar.

[0118] Step S500: Expand the fourth wind speed inversion result to the third wind speed inversion result to obtain the final joint inversion result, wherein the accuracy and observation area of ​​the final joint inversion result are consistent with the accuracy and observation area of ​​the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with the resolution of the fourth wind speed inversion result.

[0119] For step S100, in this embodiment of the application, the first wind speed inversion result can be obtained by joint inversion through the following steps S110-S140:

[0120] Step S110: Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, acquire the first observation data of the microwave scatterometer and the second observation data of the spaceborne GNSS-R.

[0121] In this embodiment of the application, the observation data of the microwave scatterometer is the radar cross section. Therefore, the first observation data can be the radar cross section of the microwave scatterometer at various locations within the observation area S1.

[0122] The observation data of spaceborne GNSS-R includes multiple types, such as bistatic radar cross section, slope of the leading edge of the time delay correlation curve, slope of the trailing edge of the time delay correlation curve, specular reflection point position, GNSS-R satellite elevation angle, scattering area, mean time delay-Doppler correlation power, and effective area.

[0123] In this embodiment of the application, the second observation data can be any two or all of the following: the mean value of time-delayed Doppler correlation power, the slope of the leading edge of the time-delayed correlation curve, and the slope of the trailing edge of the time-delayed correlation curve.

[0124] Taking the first observation data as the scattering cross-section and the second observation data as the mean value of the time-delayed Doppler correlation power, the slope of the leading edge of the time-delay correlation curve, and the slope of the trailing edge of the time-delay correlation curve, as an example, in step S110, it is necessary to obtain the scattering cross-section of the microwave scatterometer at various locations within the observation area S2, as well as the mean value of the time-delayed Doppler correlation power, the slope of the leading edge of the time-delay correlation curve, and the slope of the trailing edge of the time-delay correlation curve of the spaceborne GNSS-R within the observation area S2. The accuracy of the first observation data from the microwave scatterometer is lower than the accuracy of the second observation data from the spaceborne GNSS-R.

[0125] Step S120: Based on the geophysical model of the microwave scatterometer, obtain the first simulated observation data corresponding to the first observation data.

[0126] In this embodiment of the application, the geophysical model of the microwave scatterometer is as follows:

[0127]

[0128] Where σ is the radar backscattering coefficient, k = 0, 1, 2; W is the ground wind speed; p is the polarization mode, such as VV, HH, HV, VH; B represents relative wind speed (the angle between the observed azimuth and the wind speed); k,P(W) The harmonic coefficient can be expressed as:

[0129]

[0130] Among them, b ki,p Let be the empirical coefficient, where i is the number of terms in equation (2).

[0131] The first simulated observation data about the first data can be obtained through the geophysical model of the microwave scatterometer. For example, the first simulated observation data is the radar cross-section of each location in the observation area S2, simulated using the geophysical model of the microwave scatterometer.

[0132] Step S130: Based on the geophysical model of the spaceborne GNSS-R, obtain the second simulated observation data corresponding to the second observation data.

[0133] In this embodiment of the application, the geophysical model of the spaceborne GNSS-R is as follows:

[0134]

[0135] Where, σ 0 The cross-scattering cross section (radar cross-section) includes unknowns such as reflection coefficient and roughness to be solved; Y(τ,f) is the signal-correlated power after receiver processing, a function of delay and Doppler; P tIt is the signal power transmitted by the satellite; G t and G r It refers to the gain of the satellite antenna and receiver antenna; T i It is the coherent integration time in signal processing; R t and R r Λ(τ) is the distance from the satellite and receiver to the ground reflection point; S(f) is the Doppler frequency shift sinc function; Λ(τ) is the GNSS code correlation function; A is the effective scattering region.

[0136] Using the geophysical model of the spaceborne GNSS-R, second simulated observation data can be obtained regarding the second observation data. This second observation data includes the mean time-delay Doppler correlation power, the leading edge slope of the time-delay correlation curve, and the trailing edge slope of the time-delay correlation curve within the observation area S2. Therefore, the second simulated observation data needs to include the mean time-delay Doppler correlation power, the leading edge slope of the time-delay correlation curve, and the trailing edge slope of the time-delay correlation curve within the observation area S2, simulated using the geophysical model of the spaceborne GNSS-R.

[0137] After acquiring the observation data and simulated observation data of the microwave scatterometer and the spaceborne GNSS-R in the observation area S2, the respective observation data and simulated observation data can be used to perform joint inversion using a preset joint inversion model. That is, step S140: the first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and the preset joint inversion model.

[0138] In this embodiment of the application, the preset wind speed inversion model is as follows:

[0139]

[0140] in, This represents the result of the first wind speed inversion;

[0141] The first observation data represents the radar cross section actually observed by the microwave scatterometer at position j; σ oj For the first simulated observation data, δ represents the radar cross-section simulated by the geophysical model at the j-th location using a microwave scatterometer; j The first error value represents the uncertainty of the microwave scattering meter in measuring the radar cross section. represents the first error ratio; N indicates that at the same time in the same acquisition area (observation area S2), the microwave scatterometer acquires observation data at N different locations; k1 is the first weight corresponding to the microwave scatterometer.

[0142] DDM is the second observation data, representing DDM eigenvalues. DDM eigenvalues ​​have multiple types, such as the mean power of time-delayed Doppler correlation, the slope of the leading edge of the time-delayed correlation curve, and the slope of the trailing edge of the time-delayed correlation curve. M represents the type of DDM eigenvalue. For example, in the embodiment of this application, three DDM eigenvalues ​​are used for inversion, so M = 3. Let be the DDM characteristic value of the i-th type actually observed by the spaceborne GNSS-R within the observation area S2, and let DDM be... i The i-th type of DDM eigenvalue is simulated using the spaceborne GNSS-R geophysical model within the observation area S2; This represents the second error value for the uncertainty of the characteristic values ​​of various types of DDM measurements by spaceborne GNSS-R. The second error ratio is the sum of the error ratios of each type of DDM feature value, since there are multiple DDM feature values. k2 is the second weight corresponding to the spaceborne GNSS-R.

[0143] In this embodiment of the application, the first weight and the second weight can be preset.

[0144] In this embodiment of the application, based on the above joint inversion model (4), the wind speed inversion result can be iterated according to the preset wind speed search step size until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is used as the first wind speed inversion result.

[0145] Since the first wind speed inversion result is obtained by joint inversion based on the observation data of the microwave scatterometer and the spaceborne GNSS-R in the overlapping observation area (such as S2), the area covered by the obtained first wind speed inversion result is consistent with the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R. Figure 1 In the illustrated embodiment, the observation area of ​​the spaceborne GNSS-R is entirely within the observation area of ​​the microwave scatterometer. The overlapping observation area is the single-track observation area of ​​the spaceborne GNSS-R. Therefore, the area covered by the first wind speed inversion result can maintain the same size as the observation area of ​​the spaceborne GNSS-R. In addition, since the observation data of the spaceborne GNSS-R has high precision, the precision of the first wind speed inversion result obtained by joint inversion is consistent with the precision of the observation data of the spaceborne GNSS-R. Compared with the wind speed inversion result obtained solely based on the observation data of the microwave scatterometer, the precision is higher. Therefore, in high wind speed scenarios, the first wind speed inversion result can also have the advantage of high precision.

[0146] For step S200, based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion.

[0147] In this embodiment, the second wind speed inversion result can be obtained independently using the maximum likelihood method based solely on the observation data of the microwave scatterometer within the observation area S1. For example, the second wind speed inversion result can be obtained using the following objective function:

[0148]

[0149] Where i represents different observation directions (e.g., forward, backward) and modes (e.g., VV polarization, HV polarization, L-band, C-band), and N represents the number of observations. σ is the observed radar cross section. oi To calculate the radar cross section obtained using a geophysical model of a microwave scatterometer, δ i denoted as the standard deviation of the uncertainty in the measurement of radar cross-section by the microwave scatterometer.

[0150] For example, in this embodiment of the application, using the VV polarization observation mode, the incident angle can be determined by the objective function (5) for multiple sets of VV polarization σ0 of different frequency bands observed. Then, the vector wind field is searched and iterated along the wind speed and direction in the corresponding geophysical model. The wind speed and direction corresponding to the minimized objective function are obtained through constraint optimization, and the second wind speed inversion result is obtained at the search endpoint. The accuracy of the second wind speed inversion result is kept consistent with the accuracy of the microwave scatterometer observation data.

[0151] For step S300, the first wind speed inversion result is expanded to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer.

[0152] In this application embodiment, the expansion can be achieved through the following steps:

[0153] Step S310: Fill in the first wind speed inversion result based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result.

[0154] The second wind speed inversion result is retrieved separately based on observation data from the microwave scatterometer. Therefore, the second wind speed inversion result can cover the entire observation area S1, while the first wind speed inversion result can only cover the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, i.e., the observation area S2. In step S310, the second wind speed inversion result with a larger observation area can be used to fill in the first wind speed inversion result with a smaller observation area. During the filling process, the second wind speed inversion result remains unchanged in the observation area S2, while the portion beyond the observation area S2 can be filled using the second wind speed inversion result. The resulting filled and fused wind speed inversion result can cover the entire observation area S1. Furthermore, the filled and fused wind speed inversion result maintains high accuracy in the S2 region, while maintaining the same accuracy as the second wind speed inversion result in the region of S1 excluding S2.

[0155] Step S320: Based on the second wind speed inversion result and the filled and fused wind speed inversion result, as well as the preset first adversarial generation model, the third wind speed inversion result is obtained.

[0156] like Figure 3 As shown in the embodiment of this application, the first adversarial generative model includes a Gows generator composed of an 8-layer U-NET network. Figure 3 The system has three second wind speed inversion results and three first wind speed inversion results. By using the three second wind speed inversion results to fill in the three first wind speed inversion results, three filled and fused wind speed inversion results can be obtained.

[0157] In step S320, the second wind speed inversion result and the corresponding filled and fused wind speed inversion result can be input into the Gows generator, and the U-NET network can then generate a third wind speed inversion result. It should be noted that... Figure 3 The three Gows generators mentioned do not mean that the first adversarial generative model needs to have three generators. Figure 3 The three Gows generators are designed to correspond to the three sets of second wind speed inversion results and the infilled fused wind speed inversion results.

[0158] In this embodiment of the application, the first adversarial generative model is trained through the following steps S610-S640:

[0159] Step S610: Obtain wind speed result samples obtained by inverting observation data based on microwave scatterometers alone, and wind speed result samples after filling and fusion.

[0160] In step S610, the wind speed result samples obtained solely from the observation data of the microwave scatterometer can be acquired using the objective function (5) described above. The specific method is the same as in step S200, and will not be elaborated here. It should be noted that there can be multiple wind speed result samples obtained solely from the microwave scatterometer.

[0161] The incomplete and fused wind speed result samples can be obtained based on the wind speed result samples obtained separately from the microwave scatterometer in step S610, referring to steps 100, S200, and S310, which will not be elaborated here. It should be noted that there can be multiple incomplete and fused wind speed result samples, and each of the multiple incomplete and fused wind speed result samples corresponds one-to-one with the multiple wind speed result samples obtained separately from the microwave scatterometer in step S610.

[0162] Step S620: Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model.

[0163] Reference Figure 3 In step S620, multiple wind speed result samples obtained by inversion of individual observation data based on microwave scatterometers and multiple filled and fused wind speed result samples are input one-to-one into the first adversarial generation model to generate multiple first fusion results.

[0164] Step S630: Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively.

[0165] In step S630, the adversarial loss can be calculated by distributing the obtained first fusion result across the target observation area and matching it with the data distribution of existing wind speed products (such as ERA5 wind speed products) within that target observation area. The target observation area is the observation area reflected by the first fusion result. Specifically, it can be calculated as follows:

[0166] L GAN (G OWS D OWS ) = L OWS (G OWS D OWS (6)

[0167] Among them G OWS D represents the data distribution of the first fusion result in the target observation area. OWS For the data distribution of existing wind speed products in the target observation area, L GAN (G OWS D OWSThe first fusion result represents the adversarial loss.

[0168] The reconstruction loss can be obtained by comparing the overall structure of the first fusion result within the target observation area with the overall structure of the existing wind speed product within the same target observation area. For example, the root mean square error can be calculated pixel-by-pixel by comparing the first fusion result with the existing wind speed product (such as the ERA5 wind speed product).

[0169]

[0170] Where ows represents the wind speed value of the first fusion result generated. This represents the true wind speed of existing wind speed products, such as those obtained from real wind fields based on ERA5 wind speed products. L represents the MSE (root mean square error) loss per pixel between the first fusion result and the existing wind speed product. rec (G OWS ) represents the reconstruction loss of the first fusion result.

[0171] The allocation loss can be calculated by comparing the similarity of the feature distribution of the first fusion result and existing wind speed products in the target observation area. For example, the Kullback-Leibler (KL) divergence can be used to calculate the similarity between the probability density function of the first fusion result and existing wind speed products (such as ERA5 wind speed products) in the target observation area. Specifically, it can be calculated as follows:

[0172]

[0173] Where f(p) is the probability density function of the first fusion result in the target sensing area, g(p) is the probability density function of the existing wind speed product in the target observation area, and p represents various wind speeds, such as high wind speed or low wind speed.

[0174] Once we have obtained the adversarial loss, reconstruction loss, and assignment loss, we can optimize the first adversarial generative model.

[0175] Step S640: Optimize the first adversarial generation model based on the adversarial loss, the reconstruction loss, and the allocation loss.

[0176] In this embodiment, a loss function can be constructed using adversarial loss, reconstruction loss, and allocation loss, relating the wind speed inversion result based solely on the microwave scatterometer (the second wind speed inversion result) and the infilled fused inversion result obtained after joint inversion using the microwave scatterometer and spaceborne GNSS-R, to the third wind speed inversion result (the first fused result), as follows:

[0177] LF (G OWS D OWS ) = L GAN (G OWS D OWS )+λ1L rec (G OWS )+λ2L KL (G OWS (9)

[0178] Where λ1 is the weight of the reconstruction loss, λ2 is the weight of the allocation loss, and L F (G OWS D OWS The first adversarial generative model generates the total loss of the first fusion result. The model is trained and optimized using multiple individual wind speed samples obtained from microwave scatterometer inversion and multiple incomplete fused wind speed samples, resulting in L... F (G OWS D OWS This minimizes the probability of failure, thus ensuring that the third wind speed inversion result generated based on the first adversarial generative model has the highest possible reliability while maintaining high accuracy. Furthermore, introducing λ1 and λ2 allows control over the importance of targets and stabilizes the relative balance between training and loss.

[0179] The first wind speed inversion result is obtained by joint inversion using a microwave scatterometer and a spaceborne GNSS-R, thus giving it high accuracy, but with a small observation area. The second wind speed inversion result ensures a larger observation area. The third wind speed inversion result is obtained by expanding the first wind speed inversion result to the second wind speed inversion result. The accuracy of the third wind speed inversion result is consistent with that of the first wind speed inversion result, thus ensuring both a large observation area and high accuracy.

[0180] In another embodiment of this application, after obtaining the third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, the method further includes:

[0181] The authenticity of the third wind speed inversion result is determined through the following steps S330-S350:

[0182] Step S330: Obtain any wind speed product.

[0183] Step S340: Perform spatial matching based on the wind speed product and the third wind speed inversion result to obtain the RMSE accuracy of the third wind speed inversion result. Here, RMSE accuracy is the root mean square error.

[0184] Step S350: Determine the authenticity of the third wind speed inversion result based on the RMSE accuracy.

[0185] like Figure 3 As shown in the embodiments of this application, the first adversarial generative model further includes a first discriminator (D). OWS The first discriminator can determine the authenticity of the generated third wind speed inversion result. Specifically:

[0186] like Figure 4 As shown in this embodiment, the existing wind speed product in step S330 can be an ERA5 wind speed product ( Figure 4 (Right side) Figure 4 The left side shows the third wind speed retrieval result, where Ascat is a microwave scatterometer and cygnss is a satellite-borne GNSS-R. It should be noted that the selected ERA5 wind speed product must have the same observation area and the same observation time as the third wind speed retrieval result. For example, in... Figure 4 In the embodiment, both the third wind speed inversion result and the ERA5 wind speed product reflect the wind speed results of the target area on April 14, 2021.

[0187] After acquiring the ERA5 wind speed products for the corresponding observation area and time period, the first discriminator can average the third wind speed inversion result and the ERA5 wind speed products for that observation period and perform spatial matching. For example... Figure 5 As shown, Figure 5 That is to Figure 4 The scatter plot is obtained by averaging the third wind speed inversion results and ERA5 wind speed products within the period up to April 14, 2021, and then spatially matching them.

[0188] Then the RMSE accuracy of the third wind speed inversion result can be calculated, specifically as follows:

[0189]

[0190] Where n is the total number of data points used for accuracy verification; i = 1, 2, 3, ... represents the i-th data point; E ret,i E represents the wind speed obtained through inversion. ref,i This is the reference wind speed for the original ERA5.

[0191] After obtaining the RMSE accuracy, the authenticity of the third wind speed inversion result can be judged according to the preset RMSE accuracy threshold. If the RMSE accuracy reaches the preset threshold, proceed to the next step; otherwise, recalculation is required.

[0192] For step S400, based on the observation data of the synthetic aperture radar, a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar is obtained by inversion alone.

[0193] For example, in this embodiment of the application, wind speed inversion can be performed solely based on synthetic aperture radar (SAR) observation data. This can be achieved by utilizing existing wind speed products to provide the initial wind direction for the SAR observation area S3, and then obtaining the wind speed based on the geophysical model and cost function for the corresponding SAR band. The cost function is as follows:

[0194]

[0195] in, The normalized scattering cross section (NRCS) observed by synthetic aperture radar in the VV or HH bands. u is the preset value of the normalized scattering cross section predicted by the geophysical model. model and v model It refers to the components of wind speed in the U and V directions, obtained using standard wind speed reference values, such as those based on the ECMWF model (European Centre for Medium-Range Weather Forecasts). Δσ 0 Δu and Δv represent the Gaussian standard deviations under the ECMWF model. The geophysical model can be selected based on the band corresponding to the synthetic aperture radar. For example, in this application, the synthetic aperture radar is GF-3 (Gaofen-3 satellite), and the corresponding band is C-band, so the geophysical model of CMOD5N or CMOD7 can be selected.

[0196] The fourth wind speed inversion result can be obtained through the above cost function (10). The fourth wind speed inversion result is obtained based on the synthetic aperture radar inversion. Therefore, the resolution of the fourth wind speed inversion result can be consistent with the resolution of the synthetic aperture radar observation data, thus having the advantage of high resolution.

[0197] For step S500, the fourth wind speed inversion result is expanded to the third wind speed inversion result to obtain the final joint inversion result.

[0198] In this embodiment of the application, the final joint inversion result can be generated based on a preset second adversarial generation model, using the fourth wind speed inversion result and the third wind speed inversion result.

[0199] like Figure 6As shown in the embodiment of this application, the second adversarial generative model is GWSGAN (Global Wind SpeedGAN)-HR (High resolution), which internally has a G network composed of SRResNet. HRWS The generator inputs the fourth and third wind speed inversion results into the second adversarial generative model, and utilizes the G in the second adversarial generative model. HRWS The generator adversarially generates the final joint inversion result.

[0200] like Figure 6 As shown, the fourth wind speed inversion result obtained solely from synthetic aperture radar (SAR) has a high resolution, generally higher than that of the third wind speed inversion result sample, which typically has a resolution of 0.25°*0.25°. The SAR observation area S3 is located within the corresponding observation area S1 of the third wind speed inversion result; therefore, super-resolution can be extended based on geographic spatial similarity, utilizing G... HRWS After the generator expands the fourth wind speed inversion result to the third wind speed inversion result, it can obtain a super-resolution joint inversion result with a resolution of 0.125°*0.125°, thus ensuring that the resolution of the final joint inversion result is consistent with that of the fourth wind speed inversion result. Furthermore, the accuracy and observation area of ​​the final joint inversion result are consistent with those of the third wind speed inversion result. Therefore, the final joint wind speed inversion result not only has a large observation area and high accuracy, but also high resolution. Among them, OWSGAN-LR is the first adversarial generative model, SR 0.125° is the super-resolution joint wind speed inversion result with a resolution of 0.125°, SAR 0.125° is the SAR high-resolution wind speed inversion result with a resolution of 0.125°, RAD 0.25° is the low-resolution, high-precision microwave scatterometer wind speed inversion result with a resolution of 0.25° (the third wind speed inversion result), and conv, PReLU (Parametric Rectified Linear Unit), BN (Batch Normalization activation function), and elementwis2 SUM are the various convolutional neural networks of the generator in the second adversarial generative model.

[0201] In this embodiment of the application, the second adversarial generative model can be trained through the following steps S710-S740:

[0202] Step S710: Obtain the third wind speed inversion result sample, and the wind speed result sample obtained solely from the observation data inverted by synthetic aperture radar.

[0203] The third wind speed inversion result sample can be obtained according to steps S100-S300 above, which will not be elaborated here; the wind speed result sample obtained solely based on synthetic aperture radar inversion can be obtained by referring to step S400, which will not be elaborated here. There can be multiple third wind speed inversion result samples and multiple wind speed result samples obtained solely based on synthetic aperture radar inversion, and the accuracy of each third wind speed inversion result sample and each wind speed result sample obtained solely based on synthetic aperture radar inversion is consistent.

[0204] Step S720: Based on the wind speed result sample obtained separately from the microwave scatterometer and the third wind speed inversion result sample, generate a second fusion result using the second adversarial generative model.

[0205] In this process, multiple third-party wind speed result samples and multiple wind speed result samples obtained separately based on microwave scatterometer inversion can be input into the second adversarial generative model to generate the corresponding second fusion result.

[0206] Step S730: Calculate the adversarial loss and feature loss of the second fusion result.

[0207] In this application, the adversarial loss and feature loss of the second fusion result can be calculated in the following manner.

[0208]

[0209]

[0210] in, The feature loss is calculated pixel-by-pixel from the second fusion result and existing wind speed products (such as ERA5); To mitigate losses, a relative form is used for calculation, where x r x represents the actual wind speed of existing wind speed products. f D represents the wind speed result in the second fusion result. Ra Represents a relative discriminator, λ mse , λ Gen These represent the weights.

[0211] After obtaining the adversarial loss and feature loss of the second fusion result, the second adversarial generation model can be optimized by repeatedly generating the second fusion result and iteratively adjusting the weights of the loss function, i.e., step S740: optimize the second adversarial generation model based on the adversarial loss and the feature loss.

[0212] The optimization process of the second adversarial generative model minimizes the objective function, thereby minimizing the feature loss of the second fusion result (final wind speed inversion result) generated based on the second adversarial generative model, while ensuring high resolution.

[0213] After being trained through steps S710-S740, the second adversarial generative model can be fused with the wind speed inversion results from the higher-resolution synthetic aperture radar to the third wind speed inversion results, which have a lower resolution and a larger observation area. The resulting joint inversion results have a larger observation area, higher accuracy, and higher resolution.

[0214] In this embodiment of the application, after obtaining the final joint inversion result, the method further includes: determining the authenticity of the final joint inversion result.

[0215] like Figure 6 As shown, the second generative model also includes a second discriminator (D). HRWS The discriminator consists of several convolutional neural networks, including conv, LReLU (Leaky ReLU), BN (Batch Normalization activation function), Dense (1024) fully connected layer, Dense (1) fully connected layer, and RelativisticSigmald. The second discriminator can be used to determine the authenticity of the joint inversion result. The judgment process of the second discriminator can be referenced from the specific steps in steps S330-S350 where the first discriminator judges the authenticity of the third wind speed inversion result. These steps will not be elaborated upon here.

[0216] Based on the second discriminator, the final joint inversion result is compared with existing wind speed products (such as ERA5 wind speed products). The authenticity of the final joint inversion result is judged based on the RMSE accuracy of the joint inversion result and the preset RMSE accuracy.

[0217] The technical solution provided by this invention first performs joint inversion based on observation data from the overlapping observation area of ​​a microwave scatterometer and a spaceborne GNSS-R, resulting in a first wind speed inversion result with a relatively small observation area and high accuracy. Then, the first wind speed inversion result is expanded onto a second wind speed inversion result obtained solely from observation data based on the microwave scatterometer, resulting in a third wind speed inversion result with a larger observation area and high accuracy. Finally, a fourth wind speed inversion result obtained solely from observation data based on synthetic aperture radar is expanded onto the third wind speed inversion result, thereby obtaining a final joint inversion result with a larger observation area, higher accuracy, and higher resolution.

[0218] Exemplary device

[0219] After introducing the method of exemplary embodiments of the present invention, the following will refer to... Figure 7 An exemplary embodiment of the present invention describes a joint sea surface wind speed inversion device. This device is based on multiple observation data from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar. The observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer. The microwave scatterometer and the spaceborne GNSS-R have at least partially overlapping observation areas, and the accuracy of the observation data from the spaceborne GNSS-R is higher than that from the microwave scatterometer. The resolution of the observation data from the synthetic aperture radar is higher than that from the microwave scatterometer. The device includes:

[0220] The acquisition module is used to acquire the observation data of the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas, the observation data of the microwave scatterometer, and the observation data of the synthetic aperture radar.

[0221] The inversion module is used to jointly invert observation data from the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas to obtain a first wind speed inversion result for that area, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the spaceborne GNSS-R observation data; and

[0222] Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer.

[0223] The first adversarial generation module is used to expand the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result.

[0224] The inversion module is further configured to, based on the observation data of the synthetic aperture radar, separately invert to obtain a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar.

[0225] The second adversarial generation module is used to expand the fourth wind speed inversion result to the third wind speed inversion result to obtain a final joint inversion result, wherein the accuracy and observation area of ​​the final joint inversion result are consistent with the accuracy and observation area of ​​the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with the resolution of the fourth wind speed inversion result.

[0226] In this embodiment of the application, the acquisition module is configured as follows:

[0227] Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, first observation data from the microwave scatterometer and second observation data from the spaceborne GNSS-R are acquired; based on the geophysical model of the microwave scatterometer, first simulated observation data corresponding to the first observation data are acquired; and

[0228] Based on the geophysical model of the spaceborne GNSS-R, second simulated observation data corresponding to the second observation data is obtained;

[0229] The inversion module is configured as follows:

[0230] The first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model.

[0231] In this embodiment of the application, the acquisition module is further configured to: when acquiring the observation data of the spaceborne GNSS-R, acquire at least two of the following: the mean value of time delay Doppler correlation power, the slope of the leading edge of the time delay correlation curve, and the slope of the trailing edge of the time delay correlation curve.

[0232] In this embodiment of the application, the inversion module is configured as follows:

[0233] Based on the joint inversion model, the first error ratio of the observation data of the microwave scatterometer and the second error ratio of the observation data of the spaceborne GNSS-R are obtained respectively.

[0234] Based on the first error ratio and the second error ratio, as well as the preset weights of the microwave scatterometer observation data and the preset weights of the spaceborne GNSS-R observation data, the wind speed inversion results in the overlapping observation area of ​​the microwave scatterometer and the GNSS-R are obtained.

[0235] According to the preset wind speed search step size, the wind speed inversion result is iterated until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is taken as the first wind speed inversion result.

[0236] In this embodiment of the application, the device further includes a filling module, which is configured to:

[0237] The first wind speed inversion result is filled in based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result;

[0238] The first adversarial generation module is configured as follows:

[0239] The filled and fused wind speed inversion result is expanded to the second wind speed inversion result to obtain the third wind speed inversion result.

[0240] In this embodiment of the application, the first adversarial generative model is trained as follows:

[0241] Obtain wind speed results samples obtained from the inversion of observation data based solely on microwave scatterometers, as well as wind speed results samples after filling and fusion;

[0242] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model;

[0243] Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively;

[0244] The first adversarial generation model is optimized based on the adversarial loss, the reconstruction loss, and the allocation loss.

[0245] In this embodiment of the application, the second adversary generation module is configured as follows:

[0246] The fourth wind speed inversion result is extended to the third wind speed inversion result to obtain the final joint inversion result.

[0247] In this embodiment of the application, the second adversarial generative model is trained as follows:

[0248] Obtain the third wind speed inversion result sample, as well as the wind speed result sample obtained solely from the observation data inverted from synthetic aperture radar;

[0249] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the third wind speed inversion result sample, the second fusion result is generated using the second adversarial generative model;

[0250] Calculate the adversarial loss and feature loss of the second fusion result;

[0251] The second adversarial generation model is optimized based on the adversarial loss and the feature loss.

[0252] In this embodiment of the application, the first adversarial generative model is further used to determine the authenticity of the third wind speed inversion result after obtaining the third wind speed inversion result;

[0253] The second adversarial generative model is also used to determine the authenticity of the final joint wind speed inversion result after obtaining the final joint wind speed inversion result.

[0254] In this embodiment of the application, the first adversarial generative model is further configured as follows:

[0255] Get any wind speed product;

[0256] Spatially match the wind speed product and the third wind speed inversion result to obtain the RMSE accuracy of the third wind speed inversion result;

[0257] Based on the RMSE accuracy of the third wind speed inversion result, the authenticity of the third wind speed inversion result is determined;

[0258] The second adversarial generative model is also configured as follows:

[0259] Get any wind speed product;

[0260] Spatially match the wind speed product and the final joint wind speed inversion result to obtain the RMSE accuracy of the final joint wind speed inversion result;

[0261] The authenticity of the final joint wind speed inversion results is determined based on the RMSE accuracy of the final joint wind speed inversion results.

[0262] For the specific working principles of each module involved in the joint inversion device for sea surface wind speed, please refer to the various embodiments in the joint inversion method for sea surface wind speed, which will not be elaborated here.

[0263] The technical solution provided by this invention first performs joint inversion based on observation data from the overlapping observation area of ​​a microwave scatterometer and a spaceborne GNSS-R, resulting in a first wind speed inversion result with a relatively small observation area and high accuracy. Then, the first wind speed inversion result is expanded onto a second wind speed inversion result obtained solely from observation data from the microwave scatterometer, resulting in a third wind speed inversion result with a larger observation area and high accuracy. Finally, a fourth wind speed inversion result obtained solely from observation data from a synthetic aperture radar is expanded onto the third wind speed inversion result, thereby obtaining a final joint inversion result with a larger observation area, higher accuracy, and higher resolution.

[0264] Exemplary media

[0265] After introducing the methods and apparatus of exemplary embodiments of the present invention, the following references are made. Figure 8 A computer-readable storage medium according to an exemplary embodiment of the present invention will be described.

[0266] Please refer to Figure 8 The computer-readable storage medium shown is an optical disc 70, on which a computer program (i.e., a program product) is stored. When the computer program is run by a processor, it implements the steps described in the above-described method implementation, for example: based on the observation data of the microwave scatterometer and the spaceborne GNSS-R overlapping observation area, a first wind speed inversion result is obtained by joint inversion for the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R; based on the observation data of the microwave scatterometer, a second wind speed inversion result is obtained by separate inversion for the complete observation area of ​​the microwave scatterometer; the first wind speed inversion result is expanded to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer; based on the observation data of the synthetic aperture radar, a fourth wind speed inversion result is obtained by separate inversion for the observation area of ​​the synthetic aperture radar; the fourth wind speed inversion result is expanded to the third wind speed inversion result to obtain the final joint inversion result. The specific implementation methods of each step will not be repeated here.

[0267] It should be noted that examples of the computer-readable storage medium may also include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other optical and magnetic storage media, which will not be elaborated here.

[0268] Exemplary computing device

[0269] After introducing the methods, apparatus, and media of exemplary embodiments of the present invention, the following references are made. Figure 9 A computing device 80 according to an exemplary embodiment of the present invention will be described.

[0270] Figure 9 A block diagram is shown of an exemplary computing device 80 suitable for implementing embodiments of the present invention, which may be a computer system or a server. Figure 9 The computing device 80 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0271] like Figure 9As shown, the components of the computing device 80 may include, but are not limited to: one or more processors or processing units 801, system memory 802, and bus 803 connecting different system components (including system memory 802 and processing unit 801).

[0272] The computing device 80 typically includes a variety of computer system readable media. These media can be any available media that can be accessed by the computing device 80, including volatile and non-volatile media, removable and non-removable media.

[0273] System memory 802 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) 8021 and / or cache memory 8022. Computing device 70 may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, ROM 8023 may be used to read and write non-removable, non-volatile magnetic media (…). Figure 9 (Not shown in the image, usually referred to as "hard drive"). Although not shown in Figure 9 The diagram illustrates that disk drives for reading and writing to removable non-volatile disks (e.g., "floppy disks") and optical disc drives for reading and writing to removable non-volatile optical discs (e.g., CD-ROMs, DVD-ROMs, or other optical media) can be provided. In these cases, each drive can be connected to bus 803 via one or more data media interfaces. System memory 802 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of the present invention.

[0274] A program / utility 8025 having a set (at least one) of program modules 8024 may be stored, for example, in system memory 802, and such program modules 8024 include, but are not limited to, an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment. Program modules 8024 typically perform the functions and / or methods described in the embodiments of the present invention.

[0275] The computing device 80 can also communicate with one or more external devices 804 (such as a keyboard, pointing device, display, etc.). This communication can be performed through an input / output (I / O) interface. Furthermore, the computing device 80 can also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) via a network adapter 806. Figure 9 As shown, network adapter 806 communicates with other modules of computing device 80 (such as processing unit 801) via bus 803. It should be understood that, although... Figure 9As not shown, it can be used in conjunction with computing device 80 with other hardware and / or software modules.

[0276] The processing unit 801 executes various functional applications and data processing by running programs stored in the system memory 802. For example, based on the observation data of the microwave scatterometer and the spaceborne GNSS-R overlapping observation area, it jointly inverts to obtain a first wind speed inversion result for the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R; based on the observation data of the microwave scatterometer, it separately inverts to obtain a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer; it expands the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer; based on the observation data of the synthetic aperture radar, it separately inverts to obtain a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar; it expands the fourth wind speed inversion result to the third wind speed inversion result to obtain the final joint inversion result. The specific implementation of each step will not be repeated here. It should be noted that although several units / modules or sub-units / sub-modules of the sea surface wind speed joint inversion device are mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more units / modules described above can be embodied in one unit / module. Conversely, the features and functions of one unit / module described above can be further divided and embodied by multiple units / modules.

[0277] Furthermore, although the operations of the method of the present invention are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0278] While the spirit and principles of the invention have been described with reference to several specific embodiments, it should be understood that the invention is not limited to the disclosed specific embodiments, and the division of aspects does not imply that features in these aspects cannot be combined for benefit; such division is merely for ease of description. The invention is intended to cover various modifications and equivalent arrangements included within the spirit and scope of the appended claims.

[0279] The above description is merely a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the contents of the present invention's specification and drawings under the inventive concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

[0280] Based on the above description, the embodiments of this application provide at least the following technical solutions, but are not limited thereto:

[0281] 1. A joint inversion method for sea surface wind speed, comprising multiple observation data obtained respectively from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar, wherein the observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer, and the microwave scatterometer and the spaceborne GNSS-R have at least partially overlapping observation areas, the accuracy of the observation data from the spaceborne GNSS-R is higher than the accuracy of the observation data from the microwave scatterometer, and the resolution of the observation data from the synthetic aperture radar is higher than the resolution of the observation data from the microwave scatterometer, the method comprising:

[0282] Based on the observation data of the microwave scatterometer and the spaceborne GNSS-R within the overlapping observation area, a first wind speed inversion result is obtained by joint inversion within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the observation data of the spaceborne GNSS-R.

[0283] Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer.

[0284] The first wind speed inversion result is extended to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result;

[0285] Based on the observation data of the synthetic aperture radar, a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar is obtained by separate inversion, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar.

[0286] The fourth wind speed inversion result is extended to the third wind speed inversion result to obtain the final joint inversion result. The accuracy and observation area of ​​the final joint inversion result are consistent with those of the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with that of the fourth wind speed inversion result.

[0287] 2. The joint inversion method for sea surface wind speed as described in technical solution 1, wherein, based on the observation data of the microwave scatterometer and the spaceborne GNSS-R overlapping observation area, a first wind speed inversion result is obtained by joint inversion for the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, including:

[0288] Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, the first observation data of the microwave scatterometer and the second observation data of the spaceborne GNSS-R are acquired.

[0289] Based on the geophysical model of the microwave scatterometer, first simulated observation data corresponding to the first observation data is obtained;

[0290] Based on the geophysical model of the spaceborne GNSS-R, second simulated observation data corresponding to the second observation data is obtained;

[0291] The first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model.

[0292] 3. The joint inversion method for sea surface wind speed as described in technical solution 1 or 2, wherein the observation data of the spaceborne GNSS-R includes at least two of the following: mean value of time-delayed Doppler correlated power, slope of the leading edge of the time-delayed correlated curve, and slope of the trailing edge of the time-delayed correlated curve.

[0293] 4. The joint inversion method for sea surface wind speed as described in any one of technical solutions 1-3, wherein obtaining the first wind speed inversion result based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model includes:

[0294] Based on the joint inversion model, the first error ratio of the observation data of the microwave scatterometer and the second error ratio of the observation data of the spaceborne GNSS-R are obtained respectively.

[0295] Based on the first error ratio and the second error ratio, as well as the preset weights of the microwave scatterometer observation data and the preset weights of the spaceborne GNSS-R observation data, the wind speed inversion results in the overlapping observation area of ​​the microwave scatterometer and the GNSS-R are obtained.

[0296] According to the preset wind speed search step size, the wind speed inversion result is iterated until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is taken as the first wind speed inversion result.

[0297] 5. The joint inversion method for sea surface wind speed as described in any one of technical solutions 1-4, wherein expanding the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer includes:

[0298] The first wind speed inversion result is filled in based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result;

[0299] Based on the second wind speed inversion result and the filled and fused wind speed inversion result, as well as the preset first adversarial generation model, the third wind speed inversion result is obtained.

[0300] 6. The joint inversion method for sea surface wind speed as described in any one of technical solutions 1-5, wherein the first adversarial generative model is trained through the following steps:

[0301] Obtain wind speed results samples obtained from the inversion of observation data based solely on microwave scatterometers, as well as wind speed results samples after filling and fusion;

[0302] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model;

[0303] Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively;

[0304] The first adversarial generation model is optimized based on the adversarial loss, the reconstruction loss, and the allocation loss.

[0305] 7. The joint inversion method for sea surface wind speed as described in any one of technical solutions 1-6, wherein expanding the fourth wind speed inversion result to the third wind speed inversion result to obtain the final joint inversion result includes:

[0306] Based on the preset second adversarial generation model, the final joint inversion result is generated using the fourth wind speed inversion result and the third wind speed inversion result.

[0307] 8. The joint inversion method for sea surface wind speed as described in any one of technical solutions 1-7, wherein the second adversarial generative model is trained through the following steps:

[0308] Obtain the third wind speed inversion result sample, as well as the wind speed result sample obtained solely from the observation data inverted from synthetic aperture radar;

[0309] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the third wind speed inversion result sample, the second fusion result is generated using the second adversarial generative model;

[0310] Calculate the adversarial loss and feature loss of the second fusion result;

[0311] The second adversarial generation model is optimized based on the adversarial loss and the feature loss.

[0312] 9. The joint inversion method for sea surface wind speed as described in any one of technical solutions 1-8, wherein the method further includes:

[0313] After obtaining the third wind speed inversion result, determine the authenticity of the third wind speed inversion result; and / or

[0314] After obtaining the final joint inversion result, the authenticity of the final joint inversion result is determined.

[0315] 10. The joint inversion method for sea surface wind speed as described in any one of technical solutions 1-9, wherein the authenticity of the inversion results is determined by the following method:

[0316] Get any wind speed product;

[0317] Spatially match the wind speed product and the inversion result to obtain the RMSE accuracy of the inversion result;

[0318] The authenticity of the inversion results is determined based on the RMSE accuracy of the inversion results.

[0319] 11. A joint inversion device for sea surface wind speed, based on multiple observation data from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar, wherein the observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer, and the microwave scatterometer and the spaceborne GNSS-R have at least partially overlapping observation areas, the accuracy of the observation data from the spaceborne GNSS-R is higher than the accuracy of the observation data from the microwave scatterometer, and the resolution of the observation data from the synthetic aperture radar is higher than the resolution of the observation data from the microwave scatterometer, the device comprising:

[0320] The acquisition module is used to acquire the observation data of the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas, the observation data of the microwave scatterometer, and the observation data of the synthetic aperture radar.

[0321] The inversion module is used to jointly invert observation data from the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas to obtain a first wind speed inversion result for that area, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the spaceborne GNSS-R observation data; and

[0322] Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer.

[0323] The first adversarial generation module is used to expand the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result.

[0324] The inversion module is further configured to, based on the observation data of the synthetic aperture radar, separately invert to obtain a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar.

[0325] The second adversarial generation module is used to expand the fourth wind speed inversion result to the third wind speed inversion result to obtain a final joint inversion result, wherein the accuracy and observation area of ​​the final joint inversion result are consistent with the accuracy and observation area of ​​the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with the resolution of the fourth wind speed inversion result.

[0326] 12. The sea surface wind speed joint inversion device as described in technical solution 11, wherein the acquisition module is configured as follows:

[0327] Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, first observation data from the microwave scatterometer and second observation data from the spaceborne GNSS-R are acquired; based on the geophysical model of the microwave scatterometer, first simulated observation data corresponding to the first observation data are acquired; and

[0328] Based on the geophysical model of the spaceborne GNSS-R, second simulated observation data corresponding to the second observation data is obtained;

[0329] The inversion module is configured as follows:

[0330] The first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model.

[0331] 13. The sea surface wind speed joint inversion device as described in technical solution 11 or 12, wherein the acquisition module is further configured to: when acquiring the observation data of the spaceborne GNSS-R, acquire at least two of the following: the mean value of time delay Doppler correlated power, the slope of the leading edge of the time delay correlated curve, and the slope of the trailing edge of the time delay correlated curve.

[0332] 14. The sea surface wind speed joint inversion device as described in any one of technical solutions 11-13, wherein the inversion module is configured as follows:

[0333] Based on the joint inversion model, the first error ratio of the observation data of the microwave scatterometer and the second error ratio of the observation data of the spaceborne GNSS-R are obtained respectively.

[0334] Based on the first error ratio and the second error ratio, as well as the preset weights of the microwave scatterometer observation data and the preset weights of the spaceborne GNSS-R observation data, the wind speed inversion results in the overlapping observation area of ​​the microwave scatterometer and the GNSS-R are obtained.

[0335] According to the preset wind speed search step size, the wind speed inversion result is iterated until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is taken as the first wind speed inversion result.

[0336] 15. The sea surface wind speed joint inversion device as described in any one of technical solutions 11-14 further includes a filling module, wherein the filling module is configured as follows:

[0337] The first wind speed inversion result is filled in based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result;

[0338] The first adversarial generation module is configured as follows:

[0339] The filled and fused wind speed inversion result is expanded to the second wind speed inversion result to obtain the third wind speed inversion result.

[0340] 16. The joint inversion device for sea surface wind speed as described in any one of technical solutions 11-15, wherein the first adversarial generative model is trained as follows:

[0341] Obtain wind speed results samples obtained from the inversion of observation data based solely on microwave scatterometers, as well as wind speed results samples after filling and fusion;

[0342] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model;

[0343] Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively;

[0344] The first adversarial generation model is optimized based on the adversarial loss, the reconstruction loss, and the allocation loss.

[0345] 17. The sea surface wind speed joint inversion device as described in any one of technical solutions 11-16, wherein the second adversarial generation module is configured as follows:

[0346] The fourth wind speed inversion result is extended to the third wind speed inversion result to obtain the final joint inversion result.

[0347] 18. The joint inversion device for sea surface wind speed as described in any one of technical solutions 11-17, wherein the second adversarial generative model is trained as follows:

[0348] Obtain the third wind speed inversion result sample, as well as the wind speed result sample obtained solely from the observation data inverted from synthetic aperture radar;

[0349] Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the third wind speed inversion result sample, the second fusion result is generated using the second adversarial generative model;

[0350] Calculate the adversarial loss and feature loss of the second fusion result;

[0351] The second adversarial generation model is optimized based on the adversarial loss and the feature loss.

[0352] 19. The joint inversion device for sea surface wind speed as described in any one of technical solutions 11-18, wherein the first adversarial generative model is further used to determine the authenticity of the third wind speed inversion result after obtaining the third wind speed inversion result;

[0353] The second adversarial generative model is also used to determine the authenticity of the final joint wind speed inversion result after obtaining the final joint wind speed inversion result.

[0354] 20. The joint inversion device for sea surface wind speed as described in any one of technical solutions 11-19, wherein the first adversarial generative model is further configured as follows:

[0355] Get any wind speed product;

[0356] Spatially match the wind speed product and the third wind speed inversion result to obtain the RMSE accuracy of the third wind speed inversion result;

[0357] Based on the RMSE accuracy of the third wind speed inversion result, the authenticity of the third wind speed inversion result is determined;

[0358] The second adversarial generative model is also configured as follows:

[0359] Get any wind speed product;

[0360] Spatially match the wind speed product and the final joint wind speed inversion result to obtain the RMSE accuracy of the final joint wind speed inversion result;

[0361] The authenticity of the final joint wind speed inversion result is determined based on the RMSE accuracy of the final joint wind speed inversion result.

[0362] 21. A medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in any one of technical solutions 1-10.

[0363] 22. A computing device comprising a processor for implementing the method as described in any one of claims 1-10 when executing a computer program stored in a memory.

Claims

1. A joint inversion method for sea surface wind speed, comprising multiple observation data obtained from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar, wherein, The observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer, and the microwave scatterometer and the spaceborne GNSS-R have at least partially overlapping observation areas. The accuracy of the observation data from the spaceborne GNSS-R is higher than that from the microwave scatterometer, and the resolution of the observation data from the synthetic aperture radar is higher than that from the microwave scatterometer. The method includes: Based on the observation data of the microwave scatterometer and the spaceborne GNSS-R within the overlapping observation area, a first wind speed inversion result is obtained by joint inversion within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the observation data of the spaceborne GNSS-R. Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer. The first wind speed inversion result is extended to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result; Based on the observation data of the synthetic aperture radar, a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar is obtained by separate inversion, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar. The fourth wind speed inversion result is extended to the third wind speed inversion result to obtain the final joint inversion result. The accuracy and observation area of ​​the final joint inversion result are consistent with those of the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with that of the fourth wind speed inversion result.

2. The joint inversion method for sea surface wind speed as described in claim 1, wherein, Based on the observation data from the microwave scatterometer and the spaceborne GNSS-R overlapping observation area, a joint inversion is performed to obtain the first wind speed inversion result for the overlapping observation area, including: Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, the first observation data of the microwave scatterometer and the second observation data of the spaceborne GNSS-R are acquired. Based on the geophysical model of the microwave scatterometer, first simulated observation data corresponding to the first observation data is obtained; Based on the geophysical model of the spaceborne GNSS-R, second simulated observation data corresponding to the second observation data is obtained; The first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model.

3. The joint inversion method for sea surface wind speed as described in claim 2, wherein, The observation data of the spaceborne GNSS-R includes at least two of the following: mean time-delay Doppler correlation power, slope of the leading edge of the time-delay correlation curve, and slope of the trailing edge of the time-delay correlation curve.

4. The joint inversion method for sea surface wind speed as described in claim 2, wherein, The process of obtaining the first wind speed inversion result based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model includes: Based on the joint inversion model, the first error ratio of the observation data of the microwave scatterometer and the second error ratio of the observation data of the spaceborne GNSS-R are obtained respectively. Based on the first error ratio and the second error ratio, as well as the preset weights of the microwave scatterometer observation data and the preset weights of the spaceborne GNSS-R observation data, the wind speed inversion results in the overlapping observation area of ​​the microwave scatterometer and the GNSS-R are obtained. According to the preset wind speed search step size, the wind speed inversion result is iterated until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is taken as the first wind speed inversion result.

5. The joint inversion method for sea surface wind speed as described in claim 1, wherein, The step of expanding the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer includes: The first wind speed inversion result is filled in based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result; Based on the second wind speed inversion result and the filled and fused wind speed inversion result, as well as the preset first adversarial generation model, the third wind speed inversion result is obtained.

6. The joint inversion method for sea surface wind speed as described in claim 5, wherein, The first adversarial generative model is trained through the following steps: Obtain wind speed results samples obtained from the inversion of observation data based solely on microwave scatterometers, as well as wind speed results samples after filling and fusion; Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model; Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively; The first adversarial generation model is optimized based on the adversarial loss, the reconstruction loss, and the allocation loss.

7. The joint inversion method for sea surface wind speed as described in claim 1, wherein, The step of expanding the fourth wind speed inversion result to the third wind speed inversion result to obtain the final joint inversion result includes: Based on the preset second adversarial generation model, the final joint inversion result is generated using the fourth wind speed inversion result and the third wind speed inversion result.

8. The joint inversion method for sea surface wind speed as described in claim 7, wherein, The second adversarial generative model is trained through the following steps: Obtain the third wind speed inversion result sample, as well as the wind speed result sample obtained solely from the observation data inverted from synthetic aperture radar; Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the third wind speed inversion result sample, the second fusion result is generated using the second adversarial generative model; Calculate the adversarial loss and feature loss of the second fusion result; The second adversarial generation model is optimized based on the adversarial loss and the feature loss.

9. The joint inversion method for sea surface wind speed as described in claim 1, wherein, The method further includes: After obtaining the third wind speed inversion result, determine the authenticity of the third wind speed inversion result; and / or After obtaining the final joint inversion result, the authenticity of the final joint inversion result is determined.

10. The joint inversion method for sea surface wind speed as described in claim 9, wherein, The authenticity of the inversion results can be determined using the following method: Get any wind speed product; Spatially match the wind speed product and the inversion result to obtain the RMSE accuracy of the inversion result; The authenticity of the inversion results is determined based on the RMSE accuracy of the inversion results.

11. A joint inversion device for sea surface wind speed, based on multiple observation data from a microwave scatterometer, a spaceborne GNSS-R, and a synthetic aperture radar, wherein, The observation areas of the spaceborne GNSS-R and the synthetic aperture radar are both smaller than the observation area of ​​the microwave scatterometer, and the microwave scatterometer and the spaceborne GNSS-R have at least partially overlapping observation areas. The accuracy of the observation data from the spaceborne GNSS-R is higher than that from the microwave scatterometer, and the resolution of the observation data from the synthetic aperture radar is higher than that from the microwave scatterometer. The device includes: The acquisition module is used to acquire the observation data of the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas, the observation data of the microwave scatterometer, and the observation data of the synthetic aperture radar. The inversion module is used to jointly invert observation data from the microwave scatterometer and the spaceborne GNSS-R within their respective overlapping observation areas to obtain a first wind speed inversion result for that area, wherein the accuracy of the first wind speed inversion result is consistent with the accuracy of the spaceborne GNSS-R observation data; and Based on the observation data of the microwave scatterometer, a second wind speed inversion result for the complete observation area of ​​the microwave scatterometer is obtained by separate inversion, wherein the accuracy of the second wind speed inversion result is consistent with the accuracy of the observation data of the microwave scatterometer. The first adversarial generation module is used to expand the first wind speed inversion result to the second wind speed inversion result to obtain a third wind speed inversion result for the complete observation area of ​​the microwave scatterometer, wherein the accuracy of the third wind speed inversion result is consistent with the accuracy of the first wind speed inversion result. The inversion module is further configured to, based on the observation data of the synthetic aperture radar, separately invert to obtain a fourth wind speed inversion result for the observation area of ​​the synthetic aperture radar, wherein the resolution of the fourth wind speed inversion result is consistent with the resolution of the observation data of the synthetic aperture radar. The second adversarial generation module is used to expand the fourth wind speed inversion result to the third wind speed inversion result to obtain a final joint inversion result, wherein the accuracy and observation area of ​​the final joint inversion result are consistent with the accuracy and observation area of ​​the third wind speed inversion result, and the resolution of the final joint inversion result is consistent with the resolution of the fourth wind speed inversion result.

12. The sea surface wind speed joint inversion device as described in claim 11, wherein, The acquisition module is configured as follows: Within the overlapping observation area of ​​the microwave scatterometer and the spaceborne GNSS-R, the first observation data of the microwave scatterometer and the second observation data of the spaceborne GNSS-R are acquired. Based on the geophysical model of the microwave scatterometer, first simulated observation data corresponding to the first observation data is obtained; and Based on the geophysical model of the spaceborne GNSS-R, second simulated observation data corresponding to the second observation data is obtained; The inversion module is configured as follows: The first wind speed inversion result is obtained based on the first observation data, the second observation data, the first simulated observation data, the second simulated observation data, and a preset joint inversion model.

13. The sea surface wind speed joint inversion device as described in claim 12, wherein, The acquisition module is further configured to acquire at least two of the following when acquiring the observation data of the spaceborne GNSS-R: mean value of time delay Doppler correlation power, slope of the leading edge of the time delay correlation curve, and slope of the trailing edge of the time delay correlation curve.

14. The sea surface wind speed joint inversion device as described in claim 12, wherein, The inversion module is configured as follows: Based on the joint inversion model, the first error ratio of the observation data of the microwave scatterometer and the second error ratio of the observation data of the spaceborne GNSS-R are obtained respectively. Based on the first error ratio and the second error ratio, as well as the preset weights of the microwave scatterometer observation data and the preset weights of the spaceborne GNSS-R observation data, the wind speed inversion results in the overlapping observation area of ​​the microwave scatterometer and the GNSS-R are obtained. According to the preset wind speed search step size, the wind speed inversion result is iterated until the accuracy of the wind speed inversion result reaches the preset value, and the wind speed inversion result with the preset accuracy is taken as the first wind speed inversion result.

15. The sea surface wind speed joint inversion device as described in claim 11, further comprising a filling module, the filling module being configured to: The first wind speed inversion result is filled in based on the second wind speed inversion result to obtain the filled and fused wind speed inversion result; The first adversarial generation module is configured as follows: The filled and fused wind speed inversion result is expanded to the second wind speed inversion result to obtain the third wind speed inversion result.

16. The sea surface wind speed joint inversion device as described in claim 15, wherein, The first adversarial generative model was trained as follows: Obtain wind speed results samples obtained from the inversion of observation data based solely on microwave scatterometers, as well as wind speed results samples after filling and fusion; Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the wind speed result sample after filling and fusion, the first fusion result is generated using the first adversarial generative model; Calculate the adversarial loss, reconstruction loss, and allocation loss of the first fusion result respectively; The first adversarial generation model is optimized based on the adversarial loss, the reconstruction loss, and the allocation loss.

17. The sea surface wind speed joint inversion device as described in claim 11, wherein, The second adversarial generation module is configured as follows: The fourth wind speed inversion result is extended to the third wind speed inversion result to obtain the final joint inversion result.

18. The sea surface wind speed joint inversion device as described in claim 17, wherein, The second adversarial generative model was trained as follows: Obtain the third wind speed inversion result sample, as well as the wind speed result sample obtained solely from the observation data inverted from synthetic aperture radar; Based on the wind speed result sample obtained by inversion from the observation data based on the microwave scatterometer alone and the third wind speed inversion result sample, the second fusion result is generated using the second adversarial generative model; Calculate the adversarial loss and feature loss of the second fusion result; The second adversarial generation model is optimized based on the adversarial loss and the feature loss.

19. The sea surface wind speed joint inversion device as described in claim 11, wherein, The first adversarial generative model is also used to determine the authenticity of the third wind speed inversion result after obtaining the third wind speed inversion result; The second adversarial generative model is also used to determine the authenticity of the final joint wind speed inversion result after obtaining the final joint wind speed inversion result.

20. The sea surface wind speed joint inversion device as described in claim 11, wherein, The first adversarial generative model is also configured as follows: Get any wind speed product; Spatially match the wind speed product and the third wind speed inversion result to obtain the RMSE accuracy of the third wind speed inversion result; Based on the RMSE accuracy of the third wind speed inversion result, the authenticity of the third wind speed inversion result is determined; The second adversarial generative model is also configured as follows: Get any wind speed product; Spatially match the wind speed product and the final joint wind speed inversion result to obtain the RMSE accuracy of the final joint wind speed inversion result; The authenticity of the final joint wind speed inversion result is determined based on the RMSE accuracy of the final joint wind speed inversion result.

21. A medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in any one of claims 1-10.

22. A computing device comprising a processor for implementing the method of any one of claims 1-10 when executing a computer program stored in a memory.

Images