Spaceborne GNSS-R wave period estimation model construction method

Abstract

The invention relates to a satellite-borne GNSS-R wave period estimation model construction method. The method comprises the following steps: S1, acquiring marine environment data; s2, performing space-time matching processing on the acquired marine environment data; s3, dividing the data after space-time matching into a training set and a test set, and further dividing the training set and the test set according to the wind speed smaller than 10m / s and the wind speed larger than 10m / s to obtain a low-wind-speed sea condition data set and a high-wind-speed sea condition data set of the training set and the test set; s4, constructing and training a wave period estimation model based on the low-wind-speed sea condition data and the high-wind-speed sea condition data in the training set; and S5, carrying out wave period estimation based on the trained wave period estimation model, and verifying the performance of the model. The wave period estimation module constructed by the method can effectively improve the wave period estimation precision of high-wind-speed complex sea conditions, and an effective scheme is provided for wave period estimation by adopting a satellite-borne GNSS-R technology.

Description

Technical Field

[0001] The present invention relates to the technical field of ocean wave parameter physical model and GNSS reflection signal ocean wave period estimation, and in particular to a method for constructing a satellite-borne GNSS-R wave period estimation model. Background Art

[0002] Waves are an important part of the ocean dynamic environment. The study of wave parameters is not only crucial to marine engineering and shipping safety, but also has a profound impact on climate change, environmental protection, marine resource development and other fields. Websites such as the National Buoy Center (NDBC), the European Center for Medium-Range Weather Forecasts (ECMWF), the third-generation wave model (WaveWatchⅢ, WW3) and the National Oceanic and Atmospheric Administration (NOAA) of the United States provide data on sea level, wind, temperature, pressure and waves. However, these data have their own limitations, such as buoys, a limited number of stations, and limited coverage. ECMWF ERA5 and WW3 data have limited spatial and temporal resolution, mainly providing wave parameter data with a spatial resolution of 0.5°×0.5° and a temporal resolution of 1h, making it difficult to achieve high spatial and temporal resolution monitoring and prediction of dynamic changes in waves. In addition, the equipment currently used to monitor waves is expensive, so there are fewer stations providing wave data. In addition, satellite altimeters can provide wave height data with global coverage and climate scales, however, the spatial resolution of existing satellite altimeters is limited.

[0003] With the development of the global navigation satellite system (GNSS), spaceborne GNSS reflectometry (GNSS-R) has shown the potential to estimate ocean dynamic parameters, with significant advantages such as high temporal and spatial resolution, low cost, all-day and all-weather, and no impact from rainfall. For spaceborne GNSS-R marine applications, spaceborne GNSS-R has been used to estimate sea surface height, sea surface wind speed and direction, sea ice thickness, and significant wave height. However, to date, there are few literatures and patents on estimating significant wave height and wave period using spaceborne GNSS-R technology, and there is even no specific practical and reliable technology and method. Summary of the invention

[0004] In order to solve the current situation that there is no specific practical and reliable technology and method for estimating the significant wave height and wave period using the spaceborne GNSS-R technology, the present invention proposes a spaceborne GNSS-R wave period estimation model construction method, which uses the spaceborne GNSS-R data and ERA5 data under low wind speed and high wind speed sea conditions to propose an empirical linear model and a power function model between the wave period (T) and the significant wave height (SWH) and GNSS-R observations. ERA5 and WW3 wave period data are used to verify the performance of the empirical linear model and the power function model in estimating the wave period under low wind speed and high wind speed sea conditions. This method effectively realizes the estimation of the spaceborne GNSS-R wave period and obtains a high inversion accuracy.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for constructing a satellite-borne GNSS-R wave period estimation model, comprising:

[0007] Step S1: Acquire ocean environment data; wherein the ocean environment data includes: GNSS-R L1 level observation data, ERA5 wind speed data, significant wave height and wave period data, and third generation wave model wave period data acquired from the CYGNSS official website;

[0008] Step S2: performing time-space matching processing on the acquired marine environment data;

[0009] Step S3: Divide the data after time-space matching into a training set and a test set, and further divide the training set and the test set according to the wind speed less than 10m / s and the wind speed greater than 10m / s, respectively, to obtain a low wind speed sea condition data set and a high wind speed sea condition data set of the training set and the test set;

[0010] Step S4: constructing and training a wave period estimation model based on the low wind speed sea condition data and the high wind speed sea condition data in the training set;

[0011] Step S5: Estimating the wave period based on the trained wave period estimation model and verifying the model performance.

[0012] Optionally, step S2 includes the following sub-steps:

[0013] Step S2.1: extract the longitude and latitude of the specular reflection point, data collection time, normalized bistatic radar cross section, and front slope observation value from the GNSS-R L1 level observation data;

[0014] Step S2.2: extract the time, grid longitude and latitude, U wind component and V wind component, swell and wind wave combined significant wave height, peak wave period variables from the ERA5 data, and calculate the synthetic wind speed value at 10 m above the sea surface;

[0015] Step S2.3: extracting time, grid longitude and latitude, and peak wave period variables from WW3 data;

[0016] Step S2.4: Use the GNSS-R data collection time and GNSS-R data, ERA5 and WW3 data to perform time matching and alignment using a linear interpolation algorithm, and use the longitude and latitude information of the mirror reflection point and GNSS-R data, ERA5 and WW3 data to perform spatial matching and alignment using a bilinear interpolation algorithm.

[0017] Optionally, the calculation formula for calculating the synthetic wind speed value 10m above the sea surface is:

[0018]

[0019] Among them, u is the U wind component and v is the V wind component.

[0020] Optionally, step S3 includes the following sub-steps:

[0021] Step S3.1: Divide the data after spatiotemporal matching into a training set and a test set, which account for 60% and 40% of the total data samples respectively;

[0022] Step S3.2: further divide the training set into a data set with a wind speed less than 10 m / s and a data set with a wind speed greater than 10 m / s, to obtain a data set of low wind speed sea conditions and a data set of high wind speed sea conditions of the training set;

[0023] Step S3.3: The test set is further divided into a data set with a wind speed less than 10 m / s and a data set with a wind speed greater than 10 m / s to obtain a data set of low wind speed sea conditions and a data set of high wind speed sea conditions of the validation set.

[0024] Optionally, step S4 includes the following sub-steps:

[0025] Step S4.1: According to the wave spectrum and GNSS-R related theories, the wave period has the following relationship with the significant wave height and GNSS-R observation value:

[0026]

[0027] Where T is the wave period, SWH is the effective wave height, NBRCS is the normalized bistatic radar cross section, and LES is the leading edge slope;

[0028] Based on the relationship between the wave period, significant wave height and GNSS-R observations, the following mathematical expression is obtained:

[0029]

[0030] Among them, X is used to describe the relationship between NBRCS or LES observations and SWH and wave period;

[0031] Step S4.2: Using the low wind speed sea condition data in the training set, NBRCS and LES observations and ERA5 SWH are used to construct a wave period estimation model under low wind speed conditions; the wave period estimation model under low wind speed conditions includes: a two-parameter power function model and a three-parameter power function model. The wave period estimation model under low wind speed conditions is expressed as follows:

[0032]

[0033] Using the high wind speed sea condition data in the training set, the wave period estimation model under high wind speed conditions is constructed using NBRCS and LES observations and ERA5 SWH respectively; the wave period estimation model under high wind speed conditions includes: a linear model and a three-parameter power function model, and the wave period estimation model under high wind speed conditions is expressed as follows:

[0034]

[0035] Among them, T p is the ERA5 peak wave period, a 1 、b 1 、a 2 、b 2 、c 2 、a 3 、b 3 、a 4 、b 4 、c 4 All are model fitting parameters;

[0036] Step S4.3: First, using the low wind speed sea condition data in the training set, the fitting parameters a of the two-parameter power function model are calculated using NBRCS and LES observations and ERA5 SWH fitting. 1 and b 1 and the fitting parameter a of the three-parameter power function model 2 、b 2 、c 2 Then, the high wind speed sea condition data in the training set were used to calculate the fitting parameter a of the linear model using NBRCS and LES observations and ERA5 SWH fitting. 3 and b 3 and the fitting parameter a of the three-parameter power function model 4 、b 4 、c 4; Among them, for the linear model, the polyfit function in the MATLAB software is used to determine the model parameters, and for the power function model, the lsqcurvefit function in the MATLAB software is used to determine the model parameters.

[0037] Optionally, step S5 includes the following sub-steps:

[0038] Step S5.1: Determine the model fitting parameter a in the two-parameter power function model and the three-parameter power function model for wave period estimation using the low wind speed sea condition data set in the training set respectively. 1 、b 1 and a 2 、b 2 、c 2 After that, the wave period estimation values ​​under low wind speed conditions are calculated using the two-parameter power function model and the three-parameter power function model of wave period estimation after the parameters are determined, that is, the wave period estimation results of the two-parameter power function model and the three-parameter power function model are obtained;

[0039] Step S5.2: Determine the model fitting parameter a in the linear model for wave period estimation and the three-parameter power function model using the high wind speed sea condition data set in the training set respectively. 3 、b 3 and a 4 、b 4 、c 4 After that, the linear model of wave period estimation after determining the parameters and the three-parameter power function model are used to calculate the wave period estimation value under high wind speed conditions, that is, the wave period estimation results of the linear model and the three-parameter power function model are obtained;

[0040] Step S5.3: First, the ERA5 wave period data under low wind speed and high wind speed conditions in the test set are used as reference values, and the root mean square error, deviation, mean absolute percentage error and correlation coefficient are used to compare and analyze the performance of the wave period estimation model; then, the WW3 wave period data under low wind speed and high wind speed conditions in the test set are used as reference values, and the root mean square error, deviation, mean absolute percentage error and correlation coefficient are used to compare and analyze the performance of the wave period estimation model.

[0041] The beneficial effects of the present invention are:

[0042] Improve the accuracy of wave period estimation: By dividing the marine environment data according to wind speed conditions, the wave period estimation models under low wind speed and high wind speed conditions are trained respectively, which effectively improves the accuracy of the model's wave period estimation under different sea conditions. In particular, under high wind speed and complex sea conditions, the specially trained high wind speed wave period estimation model can more accurately capture the dynamic characteristics of sea surface fluctuations, overcoming the shortcomings of traditional methods under complex sea conditions.

[0043] Strong adaptability: The model is trained for two extreme sea conditions, low wind speed and high wind speed, respectively, which can ensure that relatively accurate wave period estimation can be obtained under different marine environmental conditions. This method is highly adaptable and can be widely used for wave period estimation in various sea areas around the world, especially in high wind speed and complex sea conditions.

[0044] Enhanced application capability of GNSS-R technology: The application of satellite-borne GNSS-R technology in wave period estimation can provide a new data acquisition method based on satellite remote sensing. Compared with traditional ocean wave observation methods (such as buoys, ship observations, etc.), GNSS-R technology has the advantage of not being restricted by sea conditions. Through the model of the present invention, GNSS signal reflection data can be used more efficiently and accurately to estimate wave period, which promotes the further development of GNSS-R technology in ocean monitoring.

[0045] Cost saving and efficiency improvement: Compared with traditional ocean observation methods, the use of satellite-borne GNSS-R technology for wave period estimation can reduce the reliance on physical equipment such as sea surface buoys and ships, thus reducing the observation cost. At the same time, the real-time and coverage of GNSS satellite data can make wave period estimation more efficient and accurate.

[0046] Improve the estimation accuracy under high wind speed sea conditions: The wave period under high wind speed varies in a complex manner, and traditional models often find it difficult to accurately estimate the wave period. The present invention, through training specifically for high wind speed sea conditions, can effectively improve the accuracy of wave period estimation under high wind speed sea conditions and meet the high-precision requirements for complex marine environments.

[0047] Promoting the development of marine environment monitoring technology: The method of the present invention not only improves the accuracy of wave period estimation, but also provides a new technical path for future marine environment monitoring. Satellite-borne GNSS-R technology can obtain global ocean wave information in real time, which has important scientific value and application significance for studying marine climate change, wave dynamics and other aspects. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0049] Figure 1 This is a flow chart of a method for constructing a satellite-borne GNSS-R wave period estimation model according to an embodiment of the present invention;

[0050] Figure 2A two-parameter power function model and a three-parameter power function model based on NBRCS and LES observation values ​​for low wind speed conditions in an embodiment of the present invention;

[0051] Figure 3 A linear model and a three-parameter power function model based on NBRCS and LES observation values ​​for high wind speed conditions of an embodiment of the present invention;

[0052] Figure 4 The comparison results of the wave period estimated by the two-parameter power function model and the three-parameter power function model based on the NBRCS and LES observation values ​​under the low wind speed conditions of the embodiment of the present invention and the ERA5 wave period data;

[0053] Figure 5 The comparison results of the wave period estimated by the linear model and the three-parameter power function model based on the NBRCS and LES observation values ​​and the ERA5 wave period data under the high wind speed conditions of the embodiment of the present invention;

[0054] Figure 6 The comparison results of the wave period estimated by the two-parameter power function model and the three-parameter power function model based on the NBRCS and LES observation values ​​under the low wind speed conditions of the embodiment of the present invention and the WW3 wave period data;

[0055] Figure 7 This is the comparison result of the wave period estimated by the linear model and three-parameter power function model based on NBRCS and LES observations under high wind speed conditions of an embodiment of the present invention and the WW3 wave period data. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] In order to verify the feasibility and reliability of the proposed method for building a satellite-borne GNSS-R wave period estimation model, the CYGNSS GNSS-R L1 V3.2 version observation data, ERA5 wind speed data, ERA5 effective wave height data, ERA5 wave period data and WW3 wave period data from April 2019 to April 2020 were obtained and downloaded. These data can be obtained free of charge from the relevant public official website. Among them, the CYGNSS GNSS-R L1 V3.2 version observation data, ERA5 effective wave height and wave period data are used to build and train the wave period estimation model, and the ERA5 wave period and WW3 wave period are used as reference data to verify the performance of the wave period estimation model. The performance verification uses four accuracy indicators: root mean square error (RMSE), bias (Bias), mean absolute percentage error (MAPE) and correlation coefficient (CC). The experimental algorithm is developed based on the MATLAB 2019b software platform.

[0059] The present embodiment proposes a method for constructing a satellite-borne GNSS-R wave period estimation model. The technical solution implementation process is as shown in the attached figure. Figure 1 As shown, the following steps are included:

[0060] Step S1, data download and acquisition, obtain GNSS-R L1 level observation data from the CYGNSS official website, and simultaneously obtain ERA5 wind speed, significant wave height and wave period data and the third generation wave model (WaveWatchⅢ, WW3) wave period data;

[0061] Step S2, data preprocessing, extracting the longitude and latitude of the specular reflection point, data collection time, normalized bistatic radar cross section (NBRCS), and leading edge slope (LES) observation value from the GNSS-R L1 level observation data, and using the longitude and latitude of the specular reflection point and the data collection time to perform time-space matching on the GNSS-R data, ERA5 data, and WW3 data respectively;

[0062] Step S3, data division, firstly divide all matching data sets into training set and test set, each accounting for 60% and 40% of the total data samples, and then further divide the training set and the test set according to the wind speed less than 10m / s and the wind speed greater than 10m / s, respectively, to obtain the low wind speed sea condition data set and the high wind speed sea condition data set of the training set and the test set;

[0063] Step S4, wave period estimation model construction and training, firstly, using the low wind speed condition data in the training set, respectively using the NBRCS and LES observations and the ERA5 effective wave height to construct the linear model and power function model of wave period estimation, then using the high wind speed condition data in the training set, respectively using the NBRCS and LES observations and the ERA5 effective wave height to construct the linear model and power function model of wave period estimation;

[0064] Step S5, wave period estimation and model performance verification, using ERA5 and WW3 wave period data as reference values, and comparing and analyzing the performance of the wave period estimation model proposed in the present invention.

[0065] As an implementation method of this embodiment, the data downloading and acquisition described in step S1, that is, obtaining GNSS-R L1 V3.2 version observation data from the CYGNSS official website; simultaneously obtaining ERA5 wind speed, including U wind component and V wind component, with a spatial resolution of 0.25°×0.25°, a temporal resolution of 1h, the observation time and grid longitude and latitude corresponding to the ERA5 wind speed; ERA5 effective wave height data (that is, the effective wave height of combined wind waves and swells) and ERA5 wave period data (that is, peak wave period) as well as observation time and grid longitude and latitude; WW3 wave period data (that is, peak wave period), WW3 observation time and grid longitude and latitude.

[0066] As an implementation of this embodiment, step S2 includes the following sub-steps:

[0067] Step S2.1, extracting the longitude and latitude of the specular reflection point, data collection time, normalized bistatic radar cross section (NBRCS), and leading edge slope (LES) observation values ​​from the GNSS-R L1 level observation data;

[0068] Step S2.2, extract the time, grid longitude and latitude, U wind component and V wind component, combined effective wave height of swell and wind wave, and peak wave period variables from the ERA5 data, and calculate the synthetic wind speed value at 10 m above the sea surface. The calculation formula is:

[0069]

[0070] Where u is the U wind component and v is the V wind component.

[0071] Step S2.3, extracting time, grid longitude and latitude, and peak wave period variables from WW3 data;

[0072] Step S2.4, use the GNSS-R data collection time and the GNSS-R data, ERA5 and WW3 data using a linear interpolation algorithm to perform time matching alignment, and then use the longitude and latitude information of the mirror reflection point and the GNSS-R data, ERA5 and WW3 data using a bilinear linear interpolation algorithm to perform spatial matching alignment.

[0073] As an implementation of this embodiment, step S3 includes the following sub-steps:

[0074] Step S3.1, divide all matching data sets into training set and test set, accounting for 60% and 40% of the total data samples respectively;

[0075] Step S3.2, further dividing the training set into a data set of low wind speed sea condition and a data set of high wind speed sea condition according to wind speed less than 10 m / s and wind speed greater than 10 m / s, to obtain a data set of low wind speed sea condition and a data set of high wind speed sea condition of the training set;

[0076] Step S3.3, further divide the validation set into a data set with a wind speed less than 10 m / s and a data set with a wind speed greater than 10 m / s, to obtain a data set with a low wind speed sea condition and a data set with a high wind speed sea condition of the validation set.

[0077] As an implementation of this embodiment, the wave period estimation model construction and training described in step 4 includes the following sub-steps:

[0078] Step S4.1, the GNSS-R NBRCS observations are related to the inverse of the long-wave mean square slope (MSS) under the geometric optics approximation, which can be expressed as:

[0079]

[0080] The slope of a wave is then dimensionally equal to the ratio L of some measure of wave height to wavelength:

[0081] slope~SWH / L(3)

[0082] Since the ocean wavelength is related to the wave period (T), the dispersion relation of deep-water gravity waves has the following relationship:

[0083]

[0084] Therefore, according to the wave spectrum and GNSS-R related theories, the wave period (T) has the following relationship with the significant wave height (SWH) and GNSS-R observations (including NBRCS and LES):

[0085]

[0086] Based on formula (5), the following mathematical expression can be obtained:

[0087]

[0088] Where X is used to describe the relationship between NBRCS (or LES) observations and SWH and wave period.

[0089] Step S4.2, using the low wind speed condition data in the training set, NBRCS and LES observations and ERA5 SWH are used to construct a two-parameter power function model and a three-parameter power function model for wave period estimation. The model expressions are as follows:

[0090]

[0091] In the formula, X is calculated by formula (6), T p is the ERA5 peak wave period, a 1 、b 1 and a 2 、b 2 、c 2 is the model fitting parameter. Among them, the first formula in formula (8) is a logarithmic expression, which can be further organized and used to calculate the model fitting parameter a using a three-parameter power function expression 2 、b 2 、c 2 .

[0092] Then, using the high wind speed condition data in the training set, the linear model and three-parameter power function model for wave period estimation were constructed using NBRCS and LES observations and ERA5SWH, respectively. The model expressions are as follows:

[0093]

[0094] In the formula, X is calculated by formula (6), T p is the ERA5 peak wave period, a 3 、b 3 and a 4 、b 4 、c 4 The first formula in formula (10) is a logarithmic expression, which can be further organized to use a three-parameter power function expression to calculate the model fitting parameter a 4 、b 4 、c 4 .

[0095] Step S4.3, firstly, using the low wind speed condition data in the training set, the fitting parameters a of the two-parameter power function model are calculated using NBRCS and LES observations and ERA5 SWH fitting. 1 and b 1 and the fitting parameter a of the three-parameter power function model 2 、b 2 、c 2 Then, the high wind speed condition data in the training set were used to calculate the fitting parameter a of the linear model using NBRCS and LES observations and ERA5 SWH fitting. 3 and b 3 and the fitting parameter a of the three-parameter power function model 4 、b 4 、c 4Specifically, for the linear model, the polyfit function in the MATLAB software was used to determine the model parameters, and for the power function model, the lsqcurvefit function in the MATLAB software was used to determine the model parameters.

[0096] Attached Figure 2 The results of fitting the two-parameter power function model and the three-parameter power function model based on the NBRCS and LES observations using the low wind speed condition data in the training set are shown. For the NBRCS observations, in the two-parameter power function model, the fitting parameter a 1 、b 1 The values ​​are: 5.09, 0.61. In the three-parameter power function model, the fitting parameter a 2 、b 2 、c 2 The values ​​are: 218.19, 0.04, -219.27; for LES observations, in the two-parameter power function model, the fitting parameter a 1 、b 1 The values ​​are: 6.02, 0.60. In the three-parameter power function model, the fitting parameter a 2 、b 2 、c 2 The values ​​are: 36.64, 0.16, -32.04 respectively.

[0097] Attached Figure 3 The results of fitting the linear model and the three-parameter power function model based on the NBRCS and LES observations using the high wind speed condition data in the training set are shown. For the NBRCS observations, in the linear model, the fitting parameter a 3 、b 3 The values ​​are: 2.32, 2.16. In the three-parameter power function model, the fitting parameter a 4 、b 4 、c 4 The values ​​are: 0.51, 1.78, 5.44; for LES observations, in the three-parameter power function model, the fitting parameter a 3 、b 3 The values ​​are: 2.75, 2.74. In the three-parameter power function model, the fitting parameter a 4 、b 4 、c 4 The values ​​are: 1.00, 1.58, and 5.29 respectively.

[0098] As an implementation of this embodiment, the wave period estimation and model performance verification described in step 5 includes the following steps:

[0099] Step S5.1, using the low wind speed condition data set in the training set, determine the model fitting parameter a in formula (7) and formula (8) respectively.1 、b 1 and a 2 、b 2 、c 2 Then, the wave period estimation values ​​under low wind speed conditions are calculated using formula (7) and formula (8), respectively, to obtain the wave period estimation results of the two-parameter power function model and the three-parameter power function model;

[0100] Step S5.2, using the high wind speed condition data set in the training set, respectively determine the model fitting parameter a in formula (9) and formula (10) 3 、b 3 and a 4 、b 4 、c 4 Then, the wave period estimation values ​​under high wind speed conditions are calculated using formula (9) and formula (10), respectively, to obtain the wave period estimation results of the linear model and the three-parameter power function model;

[0101] In step S5.3, the ERA5 wave period data under low and high wind speed conditions in the validation set are first used as reference values, and the root mean square error (RMSE), bias (Bias), mean absolute percentage error (MAPE) and correlation coefficient (CC) are used to compare and analyze the performance of the proposed wave period estimation model; then the WW3 wave period data under low and high wind speed conditions in the validation set are used as reference values, and the root mean square error (RMSE), bias (Bias), mean absolute percentage error (MAPE) and correlation coefficient (CC) are used to compare and analyze the performance of the proposed wave period estimation model.

[0102] Attached Figure 4 The results of the wave period estimation based on the two-parameter power function model and the three-parameter power function model based on NBRCS and LES observations under low wind speed conditions and the comparison with the ERA5 wave period data are shown in the Appendix. Figure 5 The results of the linear model and three-parameter power function model for estimating the wave period under high wind speed conditions based on NBRCS and LES observations and the comparison with the ERA5 wave period data are shown in Appendix. Figure 6 The results of the wave period estimation based on the two-parameter power function model and the three-parameter power function model based on the NBRCS and LES observations under low wind speed conditions and the comparison with the WW3 wave period data are shown in the Appendix. Figure 7The results of comparing the wave period estimated by the linear model and the three-parameter power function model based on NBRCS and LES observations with the WW3 wave period data under high wind speed conditions are shown. Table 1 lists the accuracy statistics of the wave period estimated by the two-parameter power function model and the three-parameter power function model based on NBRCS and LES observations under low wind speed conditions and the ERA5 wave period data, as well as the accuracy statistics of the wave period estimated by the linear model and the three-parameter power function model based on NBRCS and LES observations under high wind speed conditions and the ERA5 wave period data. Table 2 lists the accuracy statistics of the wave period estimated by the two-parameter power function model and the three-parameter power function model based on NBRCS and LES observations under low wind speed conditions and the WW3 wave period data, as well as the accuracy statistics of the wave period estimated by the linear model and the three-parameter power function model based on NBRCS and LES observations under high wind speed conditions and the WW3 wave period data.

[0103] Table 1 Accuracy statistics compared with ERA5 wave period data

[0104]

[0105] Table 2 Precision statistics compared with WW3 wave period data

[0106]

[0107]

[0108] From the attached Figure 4 -Attached Figure 7 It can be seen from Tables 1 and 2 that the wave period estimation model constructed in this embodiment based on the satellite GNSS-RNBRCS and LES observations can be used for ocean wave period estimation, among which the wave period estimation accuracy of the NBRCS observations is slightly better than that of the LES observations. Whether the ERA5 wave period data or the WW3 wave period data is used as the reference value, similar accuracy results are presented. In addition, under high wind speed conditions, the wave period accuracy estimated by the NBRCS observations and the LES observations is higher than that under low wind speed conditions. In general, the wave periods estimated by the two observations are consistent with the ERA5 and WW3 data under low and high wind speed conditions. The average absolute percentage error is less than 20.25%, which shows that the method for constructing the satellite GNSS-R wave period estimation model proposed in this embodiment is feasible and reliable.

[0109] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for constructing a satellite-borne GNSS-R wave period estimation model, characterized in that: include: Step S1: Acquire ocean environment data; wherein the ocean environment data includes: GNSS-R L1 level observation data, ERA5 wind speed data, significant wave height and wave period data, and third generation wave model wave period data acquired from the CYGNSS official website; Step S2: performing time-space matching processing on the acquired marine environment data; Step S3: Divide the data after time-space matching into a training set and a test set, and further divide the training set and the test set according to the wind speed less than 10m / s and the wind speed greater than 10m / s, respectively, to obtain a low wind speed sea condition data set and a high wind speed sea condition data set of the training set and the test set; Step S4: constructing and training a wave period estimation model based on the low wind speed sea condition data and the high wind speed sea condition data in the training set; Step S5: Estimating the wave period based on the trained wave period estimation model and verifying the model performance.

2. The method for constructing a satellite-borne GNSS-R wave period estimation model according to claim 1, characterized in that: The step S2 comprises the following sub-steps: Step S2.1: extract the longitude and latitude of the specular reflection point, data collection time, normalized bistatic radar cross section, and front slope observation value from the GNSS-R L1 level observation data; Step S2.2: extract the time, grid longitude and latitude, U wind component and V wind component, swell and wind wave combined significant wave height, peak wave period variables from the ERA5 data, and calculate the synthetic wind speed value at 10 m above the sea surface; Step S2.3: extracting time, grid longitude and latitude, and peak wave period variables from WW3 data; Step S2.4: Use the GNSS-R data collection time and GNSS-R data, ERA5 and WW3 data to perform time matching and alignment using a linear interpolation algorithm, and use the longitude and latitude information of the mirror reflection point and GNSS-R data, ERA5 and WW3 data to perform spatial matching and alignment using a bilinear interpolation algorithm.

3. The method for constructing a satellite-borne GNSS-R wave period estimation model according to claim 2, characterized in that: The calculation formula for calculating the synthetic wind speed value at 10m above the sea surface is: Among them, u is the U wind component and v is the V wind component.

4. The method for constructing a satellite-borne GNSS-R wave period estimation model according to claim 1, characterized in that: The step S3 comprises the following sub-steps: Step S3.1: Divide the data after spatiotemporal matching into a training set and a test set, which account for 60% and 40% of the total data samples respectively; Step S3.2: further divide the training set into a data set with a wind speed less than 10 m / s and a data set with a wind speed greater than 10 m / s, to obtain a data set of low wind speed sea conditions and a data set of high wind speed sea conditions of the training set; Step S3.3: The test set is further divided into a data set with a wind speed less than 10 m / s and a data set with a wind speed greater than 10 m / s to obtain a data set of low wind speed sea conditions and a data set of high wind speed sea conditions of the validation set.

5. The method for constructing a satellite-borne GNSS-R wave period estimation model according to claim 1, characterized in that: The step S4 comprises the following sub-steps: Step S4.1: According to the wave spectrum and GNSS-R related theories, the wave period has the following relationship with the significant wave height and GNSS-R observation value: Where T is the wave period, SWH is the effective wave height, NBRCS is the normalized bistatic radar cross section, and LES is the leading edge slope; Based on the relationship between the wave period, significant wave height and GNSS-R observations, the following mathematical expression is obtained: Among them, X is used to describe the relationship between NBRCS or LES observations and SWH and wave period; Step S4.2: Using the low wind speed sea condition data in the training set, NBRCS and LES observations and ERA5 SWH are used to construct a wave period estimation model under low wind speed conditions; the wave period estimation model under low wind speed conditions includes: a two-parameter power function model and a three-parameter power function model. The wave period estimation model under low wind speed conditions is expressed as follows: Using the high wind speed sea condition data in the training set, the wave period estimation model under high wind speed conditions is constructed using NBRCS and LES observations and ERA5 SWH respectively; the wave period estimation model under high wind speed conditions includes: a linear model and a three-parameter power function model, and the wave period estimation model under high wind speed conditions is expressed as follows: <h2 style=";text-align:left;direction:ltr">T<h2 style=";text-align:left;direction:ltr"> p <h2 style=";text-align:left;direction:ltr"> =a3+b3X Among them, T p is the peak wave period of ERA5, a1, b1, a2, b2, c2, a3, b3, a4, b4, c4 are all model fitting parameters; Step S4.3: First, using the low wind speed sea condition data in the training set, the NBRCS and LES observations and ERA5 SWH are used to fit and calculate the fitting parameters a1 and b1 of the two-parameter power function model and the fitting parameters a2, b2, c2 of the three-parameter power function model; then, using the high wind speed sea condition data in the training set, the NBRCS and LES observations and ERA5SWH are used to fit and calculate the fitting parameters a3 and b3 of the linear model and the fitting parameters a4, b4, c4 of the three-parameter power function model; for the linear model, the polyfit function in the MATLAB software is used to determine the model parameters, and for the power function model, the lsqcurvefit function in the MATLAB software is used to determine the model parameters.

6. The method for constructing a satellite-borne GNSS-R wave period estimation model according to claim 1, characterized in that: The step S5 comprises the following sub-steps: Step S5.1: after respectively using the low wind speed sea condition data set in the training set to determine the model fitting parameters a1, b1 and a2, b2, c2 in the two-parameter power function model and the three-parameter power function model for wave period estimation, the wave period estimation values ​​under low wind speed conditions are calculated using the two-parameter power function model and the three-parameter power function model for wave period estimation after the parameters are determined, that is, the wave period estimation results of the two-parameter power function model and the three-parameter power function model are obtained; Step S5.2: after respectively using the high wind speed sea condition data set in the training set to determine the model fitting parameters a3, b3 and a4, b4, c4 in the linear model for wave period estimation and the three-parameter power function model, the wave period estimation values ​​under high wind speed conditions are respectively calculated using the linear model for wave period estimation and the three-parameter power function model after the parameters are determined, that is, the wave period estimation results of the linear model and the three-parameter power function model are obtained; Step S5.3: First, the ERA5 wave period data under low wind speed and high wind speed conditions in the test set are used as reference values, and the root mean square error, deviation, mean absolute percentage error and correlation coefficient are used to compare and analyze the performance of the wave period estimation model; then, the WW3 wave period data under low wind speed and high wind speed conditions in the test set are used as reference values, and the root mean square error, deviation, mean absolute percentage error and correlation coefficient are used to compare and analyze the performance of the wave period estimation model.

Images