Method, device, equipment, medium and product for retrieving typhoon wind speed by synthetic aperture radar

Abstract

The application discloses a method and device for inverting typhoon wind speed by using synthetic aperture radar, equipment, medium and product, and relates to the field of ocean remote sensing and application. The method comprises the following steps: obtaining a group of observation data of the synthetic aperture radar when the synthetic aperture radar inverts the typhoon wind speed; obtaining an inversion value of the typhoon wind speed based on the weight of each observation parameter and the group of observation data of the synthetic aperture radar when the synthetic aperture radar inverts the typhoon wind speed; the determination process of the weight is as follows: constructing a target function for inverting the typhoon wind speed by using the synthetic aperture radar based on elastic network regularization; obtaining the weight of each observation parameter by solving the target function for inverting the typhoon wind speed by using the synthetic aperture radar based on elastic network regularization according to a training data set; and the training data set comprises a plurality of groups of observation data of the synthetic aperture radar and corresponding typhoon wind speeds of each group of observation data when the synthetic aperture radar monitors sample typhoon wind speed. The application can realize high-precision inversion of the typhoon wind speed.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of ocean remote sensing and application, and particularly relates to a method and device for retrieving typhoon wind speed by synthetic aperture radar, equipment, medium and product. BACKGROUND

[0002] Typhoon wind field detection methods can be divided into sea surface, air and satellite. Sea surface buoys cannot track typhoons and it is difficult to capture strong wind areas, so ships generally need to avoid the influence of typhoons for safety; aircraft carrying SFMR loads are generally used for typhoon core area detection, but the observation is difficult, the number of times is small, and it is limited to the North Atlantic and the eastern Pacific Ocean; satellite altimeter, scatterometer, radiometer and synthetic aperture radar can all obtain sea surface wind field data in a wide range and all-weather, but the altimeter can only obtain the wind speed of the ground track point, the coverage is limited, and the sea surface tends to be saturated under typhoon conditions, the scatterometer has weak high wind speed retrieval capability, the L-band radiometer can obtain high wind speed under typhoon conditions, but the spatial resolution is less than 50 km, which cannot finely depict the wind speed changes in the typhoon core area, and the synthetic aperture radar is the only satellite-borne instrument that can detect and quantitatively analyze sea surface information under extreme conditions with a spatial resolution of up to 1 kilometer, its swath can reach hundreds of kilometers and the pixel resolution can reach several meters.

[0003] In related technologies, the method for retrieving typhoon wind speed by synthetic aperture radar generally linearly fits some parameters of the synthetic aperture radar with the wind speed to predict the wind speed through the fitting curve, but the accuracy is very low. SUMMARY

[0004] The purpose of the present application is to provide a method, device, equipment, medium and product for retrieving typhoon wind speed by synthetic aperture radar, which can realize high-precision retrieval of typhoon wind speed.

[0005] To achieve the above purpose, the present application provides the following solutions:

[0006] In a first aspect, the present application provides a method for retrieving typhoon wind speed by synthetic aperture radar, comprising:

[0007] Obtaining a group of observation data of the synthetic aperture radar when the synthetic aperture radar retrieves the typhoon wind speed, the group of observation data comprising values of a plurality of observation parameters;

[0008] Based on the weight of each observation parameter and the group of observation data of the synthetic aperture radar when the synthetic aperture radar retrieves the typhoon wind speed, the retrieval value of the typhoon wind speed is obtained; wherein the determination process of the weight of each observation parameter is:

[0009] The target function of the synthetic aperture radar inversion typhoon wind speed based on the elastic network regularization is constructed; the weight of each observation parameter is obtained by solving the target function of the synthetic aperture radar inversion typhoon wind speed based on the elastic network regularization according to a training data set; the training data set comprises a plurality of groups of observation data of the synthetic aperture radar and sample typhoon wind speeds corresponding to the observation data when the synthetic aperture radar monitors the sample typhoon wind speed.

[0010] In an embodiment, each observation parameter is respectively: an incidence angle of the synthetic aperture radar, a cross-polarization backscattering coefficient, a cross-polarization equivalent noise backscattering coefficient, a vertical polarization backscattering coefficient and a vertical polarization equivalent noise backscattering coefficient.

[0011] In an embodiment, the weight of each observation parameter is obtained by solving the target function of the synthetic aperture radar inversion typhoon wind speed based on the elastic network regularization according to a training data set, and specifically comprises:

[0012] For any one observation parameter, a residual error of the observation parameter corresponding to each group of observation data in the training data set at the current iteration number is obtained according to the training data set and the weight of the observation parameter at the last iteration number.

[0013] The weight of the observation parameter at the current iteration number is calculated based on the residual error of the observation parameter corresponding to each group of observation data in the training data set at the current iteration number and a soft threshold function.

[0014] The absolute value of the difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the last iteration number is calculated.

[0015] If the absolute value of the difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the last iteration number is less than a convergence threshold, the weight of the observation parameter at the current iteration number is determined as the weight of the observation parameter.

[0016] If the absolute value of the difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the last iteration number is greater than or equal to the convergence threshold, the iteration number is updated and the next iteration is entered.

[0017] In an embodiment, the weight of the observation parameter at the current iteration number is calculated based on the residual error of the observation parameter corresponding to each group of observation data in the training data set at the current iteration number and a soft threshold function, and specifically comprises:

[0018] A first auxiliary variable of the observation parameter at the current iteration number is calculated based on the residual error of the observation parameter corresponding to each group of observation data in the training data set at the current iteration number.

[0019] calculating a second auxiliary variable of the observation parameter at the current iteration number and a third auxiliary variable of the observation parameter at the current iteration number based on the first auxiliary variable of the observation parameter at the current iteration number;

[0020] calculating a weight of the observation parameter at the current iteration number based on the second auxiliary variable of the observation parameter at the current iteration number, the third auxiliary variable of the observation parameter at the current iteration number, and a soft threshold function.

[0021] In an embodiment, the residual error of the observation parameter corresponding to each set of observation data in the training data set at the current iteration number is obtained according to the training data set and the weight of the observation parameter at the last iteration number, and specifically, the residual error of the observation parameter corresponding to each set of observation data in the training data set at the current iteration number is calculated according to the formula

[0022] If the current iteration number is the initial iteration number, the residual error of the observation parameter corresponding to each set of observation data in the training data set at the current iteration number is calculated according to the formula

[0023] If the current iteration number is not the initial iteration number, the residual error of the observation parameter corresponding to each set of observation data in the training data set at the current iteration number is calculated according to the formula

[0024] wherein, represents the residual error of the jth observation parameter corresponding to the ith set of observation data in the training data set at the initial iteration number, Y i represents the sample typhoon wind speed corresponding to the ith set of observation data in the training data set, β0represents the intercept of the inversion wind speed model, and β k represents the weight of the initial kth observation parameter, represents the value of the kth observation parameter in the ith set of observation data in the training data set, represents the residual error of the jth observation parameter corresponding to the ith set of observation data in the training data set at the n+1th iteration number, represents the residual error of the jth observation parameter corresponding to the ith set of observation data in the training data set at the nth iteration number, represents the weight of the jth observation data at the nth iteration number, represents the weight of the jth observation data at the n-1th iteration number, represents the value of the jth observation parameter in the ith set of observation data in the training data set.

[0025] In an embodiment, the weight of the observation parameter at the current iteration number is calculated based on the second auxiliary variable of the observation parameter at the current iteration number, the third auxiliary variable of the observation parameter at the current iteration number, and a soft threshold function, and specifically, the weight of the observation parameter at the current iteration number is calculated according to the formula

[0026] wherein, β j ​​​denotes the weight of the jth observation parameter at the current iteration number, SoftThreshold() denotes a soft threshold function, and γ denotes a threshold parameter. j denotes the second auxiliary variable of the observation parameter at the current iteration number, A j denotes the third auxiliary variable of the observation parameter at the current iteration number, λ denotes a regularization strength, and α denotes an elastic net mixing parameter.

[0027] In a second aspect, the present application provides a device for retrieving typhoon wind speed by synthetic aperture radar, comprising:

[0028] An acquisition module is configured to acquire a set of observation data of the synthetic aperture radar when the synthetic aperture radar is used to retrieve typhoon wind speed, the set of observation data comprising values of a plurality of observation parameters.

[0029] A retrieval module is configured to retrieve the typhoon wind speed based on the weight of each observation parameter and the set of observation data of the synthetic aperture radar when the synthetic aperture radar is used to retrieve typhoon wind speed, to obtain a retrieval value of the typhoon wind speed. The determination process of the weight of each observation parameter is as follows: a target function for retrieving typhoon wind speed by synthetic aperture radar based on elastic network regularization is constructed; the target function for retrieving typhoon wind speed by synthetic aperture radar based on elastic network regularization is solved based on a training data set to obtain the weight of each observation parameter; the training data set comprises a plurality of sets of observation data of the synthetic aperture radar when the synthetic aperture radar is used to monitor sample typhoon wind speed and the sample typhoon wind speed corresponding to each set of observation data.

[0030] In a third aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for retrieving typhoon wind speed by synthetic aperture radar according to any one of the above aspects.

[0031] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is executable on a processor to implement the method for retrieving typhoon wind speed by synthetic aperture radar according to any one of the above aspects.

[0032] In a fifth aspect, the present application provides a computer program product comprising a computer program, wherein the computer program is executable on a processor to implement the method for retrieving typhoon wind speed by synthetic aperture radar according to any one of the above aspects.

[0033] According to the embodiments provided in the present application, the present application has the following technical effects:

[0034] This application provides a method, apparatus, equipment, medium, and product for inverting typhoon wind speed using synthetic aperture radar (SAR). This application constructs a SAR objective function for inverting typhoon wind speed based on elastic network regularization; solves the SAR objective function based on elastic network regularization to obtain the weights of each observation parameter; and then obtains the inverted typhoon wind speed value based on the weights of each observation parameter, thereby improving the inversion accuracy. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 A graph showing the relationship between wind speed, wind direction, and the backscattering coefficient of a synthetic aperture radar with the same polarization;

[0037] Figure 2 Flowchart for solving the cost function;

[0038] Figure 3 Scatter plot of wind speed versus SMAP wind speed for MS1AHW2;

[0039] Figure 4 The root mean square error curves of MS1AHW2 wind speed and SMAP wind speed for different offsets are shown.

[0040] Figure 5 A scatter plot of MS1AHW2 wind speed versus SMAP wind speed after correction;

[0041] Figure 6 Scatter plot of regularized wind speed versus SMAP wind speed using an elastic network;

[0042] Figure 7 This is a flowchart of a method for retrieving typhoon wind speed using synthetic aperture radar according to an embodiment of this application;

[0043] Figure 8 This is a block diagram of a synthetic aperture radar device for retrieving typhoon wind speed according to an embodiment of this application;

[0044] Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0045] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, but not all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by those of ordinary skill in the art without creative effort belong to the scope of the present application.

[0046] The above-mentioned purposes, features and advantages of the present application will be more apparent and understandable, and the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0047] Tropical cyclones are divided into six levels, i.e. tropical depression, tropical storm, severe tropical storm, typhoon, strong typhoon and super strong typhoon, according to the maximum average wind speed near the center of the lower layer. The "typhoon" described in the present application refers to the tropical cyclone with a maximum average wind speed near the center of the lower layer greater than or equal to 24.5 m / s.

[0048] Table 1: Tropical cyclone levels

[0049] Tropical cyclone category Wind speed category Wind speed value (m / s) Tropical depression 6～7 10.8～17.1 Tropical storm 8～9 17.2～24.4 Strong tropical storm 10～11 24.5～32.6 Typhoon 12～13 32.7～41.4 Strong typhoon 14～15 41.5～50.9 Super typhoon ≥16 ≥51.0

[0050] The backscattering coefficient of the co-polarization (vertical polarization transmission and vertical polarization reception, or horizontal polarization transmission and horizontal polarization reception) synthetic aperture radar changes with the increase of the wind direction value as shown in part (a) of FIG. 1, and increases with the increase of the wind speed value as shown in part (b) of FIG. 1. The greater the wind speed value, the smaller the change of the backscattering coefficient. Especially under typhoon conditions (the maximum average wind speed near the center of the lower layer is greater than or equal to 24.5 m / s), the backscattering coefficient of the co-polarization synthetic aperture radar is not sensitive to the wind speed. Figure 1 Figure 1 In recent years, studies have shown that the backscattering coefficient of the cross-polarization (vertical polarization transmission and horizontal polarization reception, or horizontal polarization transmission and vertical polarization reception) synthetic aperture radar presents completely different characteristics from the backscattering coefficient of the co-polarization synthetic aperture radar. The backscattering coefficient of the cross-polarization synthetic aperture radar increases with the increase of the wind speed value, and is not sensitive to the wind direction value. Further studies have shown that the equivalent noise coefficient of the synthetic aperture radar and the incident angle have an important influence on the inversion of the wind speed.

[0051] Based on this, in an exemplary embodiment, the present application provides a method for inverting the wind speed of a typhoon by a synthetic aperture radar, as shown in FIG. 2, specifically comprising the following steps.

[0052] Figure 7

[0053] Step 101: obtaining a set of observation data of the synthetic aperture radar when inverting the wind speed of a typhoon by the synthetic aperture radar, the set of observation data including values of a plurality of observation parameters.​​​

[0054] Step 102: based on the weight of each observation parameter and the inversion of the typhoon wind speed by the synthetic aperture radar, a set of observation data of the synthetic aperture radar obtains the inversion value of the typhoon wind speed; wherein the determination process of the weight of each observation parameter is: constructing a synthetic aperture radar typhoon wind speed inversion target function based on elastic network regularization; obtaining the weight of each observation parameter by solving the synthetic aperture radar typhoon wind speed inversion target function based on elastic network regularization according to the training data set; the training data set includes multiple sets of observation data of the synthetic aperture radar and the corresponding sample typhoon wind speed of each set of observation data when the synthetic aperture radar monitors the sample typhoon wind speed.

[0055] In another exemplary embodiment of the present application, each observation parameter is respectively: synthetic aperture radar incidence angle, cross-polarization backscattering coefficient, cross-polarization equivalent noise backscattering coefficient, vertical polarization backscattering coefficient and vertical polarization equivalent noise backscattering coefficient. The present application comprehensively considers the synthetic aperture radar incidence angle, cross-polarization backscattering coefficient, cross-polarization equivalent noise backscattering coefficient, vertical polarization backscattering coefficient and vertical polarization equivalent noise backscattering coefficient, constructs a synthetic aperture radar typhoon wind speed inversion target function based on elastic network regularization, gives a calculation implementation method, and realizes high-precision inversion of cross-polarization synthetic aperture radar typhoon wind speed.

[0056] In another exemplary embodiment of the present application, the synthetic aperture radar typhoon wind speed inversion target function based on elastic network regularization is constructed, and the specific process is:

[0057] Suppose the observation data is X=[X1, X2, X3, X4, X5], which are respectively the synthetic aperture radar incidence angle X1, the cross-polarization backscattering coefficient X2, the cross-polarization equivalent noise backscattering coefficient X3, the vertical polarization backscattering coefficient X4 and the vertical polarization equivalent noise backscattering coefficient X5, and the output parameter is the sea surface wind speed (i.e. typhoon wind speed) Y. Assuming that the inversion wind speed model is linear:

[0058] Y=β0+β1X1+β2X2+β3X3+β4X4+β5X5 (1)

[0059] Wherein, β0represents the intercept of the model, β j represents the weight of the jth observation parameter (j=1,...,5), and the initial value is 0.

[0060] The model target is to minimize the elastic network regularization target function: J(β)=loss function+regularization term.

[0061] Let the loss function be:

[0062]

[0063] m is the number of groups of observation data in the training data set, y i represents the sample typhoon wind speed corresponding to the i-th group of observation data in the training data set, represents the value of the j-th observation parameter in the i-th group of observation data in the training data set.

[0064] The regularization term is a linear combination of L1 regularization and L2 regularization ||β||1, That is

[0065]

[0066] where λ represents the regularization strength. α represents the elastic network mixing parameter, which is used to adjust the weight of L1 regularization and L2 regularization (α=1 for Lasso, α=0 for Ridge).

[0067] Therefore, the synthetic aperture radar typhoon wind speed inversion objective function based on elastic network regularization is:

[0068] J(β)=L(β)+R(β) (4)

[0069] In another exemplary embodiment of the present application, based on the weight of each observation parameter and the inversion value of the typhoon wind speed obtained by the synthetic aperture radar when the synthetic aperture radar is used to invert the typhoon wind speed, the weight of each observation parameter is obtained by substituting the weight of each observation parameter and the inversion value of the typhoon wind speed obtained by the synthetic aperture radar when the synthetic aperture radar is used to invert the typhoon wind speed into formula (1).

[0070] In another exemplary embodiment of the present application, the weight of each observation parameter is obtained by solving the synthetic aperture radar typhoon wind speed inversion objective function based on elastic network regularization according to the training data set, which specifically includes:

[0071] For any observation parameter, according to the training data set and the weight of the observation parameter in the last iteration, the residual of the observation parameter corresponding to each group of observation data in the training data set in the current iteration is obtained.

[0072] Based on the residual of the observation parameter corresponding to each group of observation data in the training data set in the current iteration and the soft threshold function, the weight of the observation parameter in the current iteration is calculated.

[0073] The absolute value of the difference between the weight of the observation parameter in the current iteration and the weight of the observation parameter in the last iteration is calculated.

[0074] If the absolute value of the difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the last iteration number is less than the convergence threshold, the weight of the observation parameter at the current iteration number is determined as the weight of the observation parameter.

[0075] If the absolute value of the difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the last iteration number is greater than or equal to the convergence threshold, the iteration number is updated to enter the next iteration.

[0076] In another exemplary embodiment of the present application, the weight of the observation parameter at the current iteration number is calculated based on the residual of the observation parameter corresponding to each group of observation data in the training data set at the current iteration number and the soft threshold function, and specifically includes:

[0077] The first auxiliary variable of the observation parameter at the current iteration number is calculated based on the residual of the observation parameter corresponding to each group of observation data in the training data set at the current iteration number.

[0078] The second auxiliary variable of the observation parameter at the current iteration number and the third auxiliary variable of the observation parameter at the current iteration number are calculated based on the first auxiliary variable of the observation parameter at the current iteration number.

[0079] The weight of the observation parameter at the current iteration number is calculated based on the second auxiliary variable of the observation parameter at the current iteration number, the third auxiliary variable of the observation parameter at the current iteration number and the soft threshold function.

[0080] In practical applications, because the elastic network has no analytical solution, it is assumed that all parameters are fixed values except β j (j = 1,..., 5), and only β j is a variable, and an iterative method is used to solve β j .

[0081] Let the residual of the jth observation parameter β j corresponding to the ith group of observation data in the training data set be r i

[0082]

[0083] Therefore, the objective function can be simplified to only contain the term of β j :

[0084]

[0085] Expanding and combining like terms, we get:

[0086]

[0087] The derivative of β j is calculated and set to zero to obtain:

[0088]

[0089] where sign() represents the sign function.

[0090] Define auxiliary variables:

[0091] The first auxiliary variable of the jth observation parameter β j is:

[0092]

[0093] The second auxiliary variable of the jth observation parameter β j is:

[0094]

[0095] The third auxiliary variable of the jth observation parameter β j is:

[0096]

[0097] The soft threshold form is:

[0098]

[0099] where the soft threshold function is defined as:

[0100]

[0101] The initial value of β0is calculated according to the formula , where Y i represents the sample typhoon wind speed corresponding to the ith set of observation data in the training data set, m is the number of observation data sets in the training data set, and the maximum number of iterations is set to N (N = 100), the current iteration number is n, and the convergence threshold is δ (δ = 0.001).

[0102] If

[0103]

[0104] Let

[0105]

[0106] Enter the next iteration until

[0107]

[0108] Output the current β j and perform the next iteration to solve for β j .

[0109] After obtaining β1, β2, β3, β4 and β5, β0 is obtained

[0110]

[0111] Thus, the required β0, β1, β2, β3, β4 and β5 are obtained.

[0112] The application also provides an embodiment to illustrate the effect of the wind speed inversion method provided in the above embodiment, and the specific steps include:

[0113] Step 5.1: Collect Sentinel-1 satellite C-band synthetic aperture radar data and SMAP satellite L-band radiometer sea surface wind speed inversion product data, carry out spatial and temporal matching, and obtain a matched data set.

[0114] Step 5.1.1: Sentinel-1 synthetic aperture radar.

[0115] Sentinel-1 satellite carries C-band (5.405 GHz) synthetic aperture radar, operates in the same orbit as the sun, and detects oceans and land in four imaging modes of SM (80 km wide, 5 m x 5 m resolution), IW (250 km wide, 5 m x 20 m), EW (400 km wide, 25 m x 100 m resolution) and WV (100 km wide, 5 m x 20 m resolution). Sentinel-1A was launched on April 3, 2014. Sentinel-1B was launched on April 25, 2016. The spacecraft experienced an anomaly related to the instrument electronics power supply provided by the satellite platform, resulting in the inability to provide radar data since December 23, 2021, so the European Space Agency and the European Commission announced the end of the Sentinel-1B mission on August 3, 2022. Sentinel-1C was launched on December 5, 2024, replacing Sentinel-1B. Sentinel-1 satellites can be used to detect sea surface wind fields, with an average resolution of 1 km x 1 km.

[0116] Step 5.1.2: Soil Moisture Active Passive Satellite

[0117] The Soil Moisture Active Passive (SMAP) satellite is in a 685 km altitude near-polar orbit with an inclination of 98 degrees and an ascending node time of 18:00, achieving global coverage in about 3 days with a precise 8-day revisit period. The satellite consists of an active radar and a passive radiometer sharing the same L-band feed horn. The satellite was launched on January 31, 2015, and the passive radiometer has been operating normally since then with a resolution of about 39 km x 47 km. The active radar failed on July 7, 2015. The main task of SMAP is to measure soil moisture, and it can also measure sea surface salinity and sea surface wind speed. From 18 m / s to 70 m / s wind speed range, the microwave radiation received by SMAP increases with the increase of wind speed and is almost not affected by rainfall. Especially in the wind speed range above 25 m / s, the deviation of SMAP wind speed compared with the Stepped Frequency Microwave Radiometer (SFMR) is only 0.5 m / s, and the standard deviation is only 3.01 m / s. The data is obtained from www.remss.com, and the sea surface wind field product data has been interpolated into a 0.25° and 0.25° resolution latitude-longitude grid.

[0118] Step 5.1.3: Data preprocessing.

[0119] The spatial distance is not greater than 0.25° and the time interval is not greater than 3h as the criterion for spatio-temporal matching, and the incidence angle of synthetic aperture radar, cross-polarization backscatter coefficient, cross-polarization equivalent noise backscatter coefficient, vertical polarization backscatter coefficient and vertical polarization equivalent noise backscatter coefficient and SMAP satellite wind speed are extracted, obtaining 75750 matched data pairs. Among them, there are 5451 matched data pairs in the SMAP satellite wind speed product with wind speed greater than or equal to 24.5 m / s. Randomly select 3816 matched data pairs as the training data set, and the remaining 1635 matched data pairs as the test data set.

[0120] Step 5.2: Cost function solution.

[0121] Let the maximum number of iterations be N = 100, the convergence threshold be δ = 0.001, the initial value of iteration be β0, which is the average of 3816 typhoon wind speeds in the training data set, [β1, β2, β3, β4, β5] = [0, 0, 0, 0, 0], the regularization strength λ = 0.1, and the elastic network hybrid parameter α = 0.5. The cost function solution process is shown in Figure 2

[0122] ​First step, let β1 be a variable, and the rest of the parameters are fixed values, the synthetic aperture radar incidence angle, cross-polarization backscatter coefficient, cross-polarization equivalent noise backscatter coefficient, vertical polarization backscatter coefficient and vertical polarization equivalent noise backscatter coefficient in the training data set are taken as five characteristic quantities to form a set of observation data X = [X1, X2, X3, X4, X5], and the SMAP wind speed is taken as the output parameter y, which is substituted into formula (5) to obtain the parameter residual.

[0123] Second step, substitute the calculated parameter residual, training data set, regularization strength and elastic network hybrid parameter into formula (9) to formula (11) to obtain the first auxiliary variable z1, the second auxiliary variable γ1 and the third auxiliary variable A1.

[0124] Third step, substitute z1, γ1 and A1 into formula (12), and use the soft threshold function of formula (13) to obtain the weight under this round of iteration

[0125] Fourth step, compare and If , use formula (15) to calculate the new parameter residual, and go to the second step; if , then is the optimal parameter weight and is output.

[0126] Step 6: result comparison.

[0127] Step 6.1: At present, the model used by the Copernicus Marine Service Center of the European Space Agency for Sentinel-1A synthetic aperture radar typhoon wind speed retrieval is the cross-polarization geophysical model (MS1AHW2), and the expression is:

[0128]

[0129] Where, σ vh represents the cross-polarization backscatter coefficient of synthetic aperture radar, Y represents the typhoon wind speed, a z1 , b z1 , a z2 , b z2 are intermediate variables.

[0130]

[0131] The specific coefficient values are shown in Table 2.

[0132] Table 2 Coefficient values of cross-polarization geophysical model (MS1AHW2)

[0133] Coefficient Value Coefficient Value Coefficient Value a 0,z1 ]]> 2.13755392e-06 a 2,z2 ]]> 2.87698338e-08 [cd] -0.23257086 b 0,z1 ]]> 2.47395267 b 0,z2 ]]> 1.14509104 ​ 12.39717002 b 1,z1 ]]> -2.85775085e-03 b 1,z2 ]]> 3.41828829e-02 [ca2] 0.21667263 a 0,z2 ]]> 6.54058552e-05 b 2,z2 ]]> -4.79715441e-04 [c3] 12.22862991 a 1,z2 ]]> -2.43845137e-06

[0134] The synthetic aperture radar incidence angle and backscatter coefficient in the test data set are directly substituted into the cross-polarized geophysical model (MS1AHW2) to obtain the MS1AHW2 wind speed V MS1AHW2 As shown in the SMAP wind speed scatter plot in the test data set above Figure 3 , the MS1AHW2 wind speed is obviously systematically overestimated, and the root mean square error between the MS1AHW2 wind speed and the SMAP wind speed in the test data set is 5.18 m / s.

[0135] The above inversion result is subtracted by the offset ∈ [0, 7], and the specific formula is Y offset_MS1AHW2 = Y MS1AHW2 -offset, and the root mean square error of the result is shown in Figure 4 When offset = 3.7, the root mean square error between the MS1AHW2 wind speed and the SMAP wind speed is the smallest, which is 3.65 m / s.

[0136] Step 6.2: Substitute the synthetic aperture radar incidence angle, cross-polarized backscatter coefficient, cross-polarized equivalent noise backscatter coefficient, vertical polarization backscatter coefficient, vertical polarization equivalent noise backscatter coefficient, and optimal weight in the test data set above into formula (1), and the output result of the elastic network regularization model can be obtained. The elastic network regularization wind speed and the SMAP wind speed scatter plot in the test data set are shown in Figure 6 , and compared with the SMAP satellite wind speed in the test data set, the wind speed error is 3.39 m / s.

[0137] As shown in Table 3, the root mean square error between the European Space Agency MS1AHW2 wind speed and the SMAP wind speed is 5.18 m / s, which is obviously systematically overestimated. After bias correction, the MS1AHW2 wind speed and the SMAP wind speed scatter plot after correction are shown in Figure 5 , and the root mean square error between the MS1AHW2 wind speed and the SMAP wind speed after correction is reduced to 3.65 m / s. After elastic network regularization training, the wind speed inversion error is 3.39 m / s, which is obviously better than the European Space Agency MS1AHW2 wind speed accuracy.

[0138] Table 3 Comparison of MS1AHW2 wind speed and elastic network regularization wind speed error

[0139]

[0140] Based on the same inventive concept, the application further provides a device for implementing the method for retrieving typhoon wind speed by synthetic aperture radar. The device provides a solution similar to the implementation solution described in the above method, so the specific limitations in one or more device embodiments for retrieving typhoon wind speed by synthetic aperture radar provided below can refer to the limitations of the method for retrieving typhoon wind speed by synthetic aperture radar described above, and will not be repeated here.

[0141] In an exemplary embodiment, a device for retrieving typhoon wind speed by synthetic aperture radar is provided, as shown in Figure 8 comprises:

[0142] The acquisition module A1 is configured to acquire a set of observation data of the synthetic aperture radar when the synthetic aperture radar retrieves the typhoon wind speed, and the set of observation data comprises values of a plurality of observation parameters.

[0143] The retrieval module A2 is configured to obtain a retrieval value of the typhoon wind speed based on the weight of each observation parameter and the set of observation data of the synthetic aperture radar when the synthetic aperture radar retrieves the typhoon wind speed. The determination process of the weight of each observation parameter is as follows: constructing an elastic network regularization-based synthetic aperture radar retrieval typhoon wind speed objective function; obtaining the weight of each observation parameter by solving the elastic network regularization-based synthetic aperture radar retrieval typhoon wind speed objective function according to a training data set. The training data set comprises a plurality of sets of observation data of the synthetic aperture radar when the synthetic aperture radar monitors sample typhoon wind speed and the sample typhoon wind speed corresponding to each set of observation data.

[0144] In an exemplary embodiment, a computer device is provided, which can be a server or a terminal, and its internal structure diagram can be as shown in Figure 9 The computer device comprises a processor, a memory, an input / output interface (I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store synthetic aperture radar retrieval typhoon wind speed data. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement a method for retrieving typhoon wind speed by synthetic aperture radar.

[0145] Those skilled in the art can understand that Figure 9 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, which includes a memory and a processor. The memory stores a computer program. The processor executes the computer program to implement the above-mentioned method embodiments.

[0146] In an exemplary embodiment, a computer readable storage medium is provided, which stores a computer program. The computer program is executed by a processor to implement the above-mentioned method embodiments.

[0147] In an exemplary embodiment, a computer program product is provided, which includes a computer program. The computer program is executed by a processor to implement the above-mentioned method embodiments.

[0148] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0149] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, databases or other media used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc.

[0150] The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a blockchain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0151] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, all possible combinations of the technical features in the above embodiments are not described, but as long as the combinations of the technical features do not exist, they should be considered as the scope of the present application.

[0152] The principles and implementation modes of the present application are described by applying specific examples herein. The above description of the embodiments is only used to help understand the method and its core idea of the present application; meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range can be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A method for retrieving typhoon wind speed from synthetic aperture radar, characterized in that, The method for inverting the wind speed of a typhoon by the synthetic aperture radar comprises the following steps: A set of observation data of the synthetic aperture radar is obtained when the synthetic aperture radar inverts the wind speed of a typhoon, and the set of observation data comprises values of multiple observation parameters; The inversion value of the wind speed of a typhoon is obtained based on the weight of each observation parameter and the set of observation data of the synthetic aperture radar when the synthetic aperture radar inverts the wind speed of a typhoon; wherein the determination process of the weight of each observation parameter is as follows: A target function for inverting the wind speed of a typhoon by the synthetic aperture radar based on elastic network regularization is constructed; and the target function is solved based on a training data set to obtain the weight of each observation parameter, specifically including the following steps: For any observation parameter, the residual of the observation parameter corresponding to each set of observation data in the training data set at the current iteration is obtained based on the training data set and the weight of the observation parameter at the last iteration; A first auxiliary variable of the observation parameter at the current iteration is calculated based on the residual of the observation parameter corresponding to each set of observation data in the training data set at the current iteration; A second auxiliary variable and a third auxiliary variable of the observation parameter at the current iteration are calculated based on the first auxiliary variable of the observation parameter at the current iteration; The weight of the observation parameter at the current iteration is calculated based on the second auxiliary variable of the observation parameter at the current iteration, the third auxiliary variable of the observation parameter at the current iteration and a soft threshold function; The absolute value of the difference between the weight of the observation parameter at the current iteration and the weight of the observation parameter at the last iteration is calculated; If the absolute value of the difference between the weight of the observation parameter at the current iteration and the weight of the observation parameter at the last iteration is less than a convergence threshold, the weight of the observation parameter at the current iteration is determined as the weight of the observation parameter; If the absolute value of the difference between the weight of the observation parameter at the current iteration and the weight of the observation parameter at the last iteration is greater than or equal to the convergence threshold, the iteration number is updated, and the next iteration is entered; the training data set comprises multiple sets of observation data of the synthetic aperture radar and the sample typhoon wind speed corresponding to each set of observation data when the synthetic aperture radar monitors the sample typhoon wind speed.

2. The method of claim 1, wherein, Each observation parameter is respectively the incidence angle of the synthetic aperture radar, the cross-polarization backscattering coefficient, the cross-polarization equivalent noise backscattering coefficient, the vertical polarization backscattering coefficient and the vertical polarization equivalent noise backscattering coefficient.

3. The method of claim 1, wherein, The residual of the observation parameter corresponding to each set of observation data in the training data set at the current iteration is obtained based on the training data set and the weight of the observation parameter at the last iteration, specifically including the following steps: If the current iteration number is the initial iteration number, then the formula is calculated. If the current iteration number is not the initial iteration number, then the formula is calculated. wherein, represents the residual of the jth observation parameter corresponding to the ith set of observation data in the training dataset at the initial iteration number, represents the sample typhoon wind speed corresponding to the ith set of observation data in the training dataset, represents the intercept of the inversion wind speed model, represents the initial weight of the kth observation parameter, represents the value of the kth observation parameter in the ith set of observation data in the training dataset, represents the residual of the jth observation parameter corresponding to the ith set of observation data in the training dataset at the n+1th iteration number, represents the residual of the jth observation parameter corresponding to the ith set of observation data in the training dataset at the nth iteration number, represents the weight of the jth observation data at the nth iteration number, represents the weight of the jth observation data at the n-1th iteration number, represents the value of the jth observation parameter in the ith set of observation data in the training dataset.

4. The method of claim 1, wherein, The weight of the observation parameter at the current iteration is calculated based on the second auxiliary variable of the observation parameter at the current iteration, the third auxiliary variable of the observation parameter at the current iteration and a soft threshold function, specifically including the following steps: According to the formula is calculated, wherein denotes the weight of the jth observed parameter at the current iteration number, denotes the soft threshold function, denotes the second auxiliary variable of the observed parameter at the current iteration number, denotes the third auxiliary variable of the observed parameter at the current iteration number, denotes the regularization strength, denotes the elastic net mixing parameter.

5. An apparatus for retrieving typhoon wind speed using synthetic aperture radar, characterized in that, The device for inverting the wind speed of a typhoon by the synthetic aperture radar comprises The acquisition module is configured to acquire a set of observation data of the synthetic aperture radar when the synthetic aperture radar is used to retrieve a typhoon wind speed, the set of observation data including values of a plurality of observation parameters; The retrieval module is configured to obtain a retrieval value of the typhoon wind speed based on the weights of the observation parameters and the set of observation data of the synthetic aperture radar when the synthetic aperture radar is used to retrieve the typhoon wind speed; and wherein the weights of the observation parameters are determined by constructing an elastic network regularization-based synthetic aperture radar typhoon wind speed retrieval objective function, and solving the elastic network regularization-based synthetic aperture radar typhoon wind speed retrieval objective function based on a training data set to obtain the weights of the observation parameters, specifically including: For any observation parameter, at a current iteration number, residual errors of the observation parameter corresponding to each set of observation data in the training data set at the current iteration number are obtained based on the training data set and the weight of the observation parameter at a previous iteration number; A first auxiliary variable of the observation parameter at the current iteration number is calculated based on the residual errors of the observation parameter corresponding to each set of observation data in the training data set at the current iteration number; A second auxiliary variable of the observation parameter at the current iteration number and a third auxiliary variable of the observation parameter at the current iteration number are calculated based on the first auxiliary variable of the observation parameter at the current iteration number; The weight of the observation parameter at the current iteration number is calculated based on the second auxiliary variable of the observation parameter at the current iteration number, the third auxiliary variable of the observation parameter at the current iteration number, and a soft threshold function; An absolute value of a difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the previous iteration number is calculated; If the absolute value of the difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the previous iteration number is less than a convergence threshold, the weight of the observation parameter at the current iteration number is determined as the weight of the observation parameter; If the absolute value of the difference between the weight of the observation parameter at the current iteration number and the weight of the observation parameter at the previous iteration number is greater than or equal to the convergence threshold, the iteration number is updated, and the next iteration is entered; and the training data set includes a plurality of sets of observation data of the synthetic aperture radar and sample typhoon wind speeds corresponding to the sets of observation data when the synthetic aperture radar is used to monitor sample typhoon wind speeds.

6. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for retrieving a typhoon wind speed of a synthetic aperture radar according to any one of claims 1-4.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the method for retrieving a typhoon wind speed of a synthetic aperture radar according to any one of claims 1-4.

8. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the method for retrieving a typhoon wind speed of a synthetic aperture radar according to any one of claims 1-4.

Images