Home Journals Articles Updates About
CN
image
收藏切换
Validation of significant wave height retrieval from co-polarization Chinese Gaofen-3 SAR imagery using an improved algorithm
收藏切换
PDF
Yexin SHENG1, Weizeng SHAO1, *, Shuai ZHU1, Jian SUN2, Xinzhe YUAN3, Shuiqing LI4, 5, Jian SHI6, Juncheng ZUO1
Acta Oceanologica Sinica | 2018, 37(6) : 1 - 10
Less
收藏切换
Acta Oceanologica Sinica | 2018, 37(6): 1-10
• Physical Oceanography,Marine Meteorology and Marine Physics •
Validation of significant wave height retrieval from co-polarization Chinese Gaofen-3 SAR imagery using an improved algorithm
Full
Yexin SHENG1, Weizeng SHAO1, *, Shuai ZHU1, Jian SUN2, Xinzhe YUAN3, Shuiqing LI4, 5, Jian SHI6, Juncheng ZUO1
Affiliations
  • 1 Marine Science and Technology College, Zhejiang Ocean University, Zhoushan 316000, China
  • 2 Physical Oceanography Laboratory/CIMST, Ocean University of China and Qingdao National Laboratory for Marine Science and Technology, Qingdao 266100, China
  • 3 National Satellite Ocean Application Service, State Oceanic Administration, Beijing 100081, China
  • 4 Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071, China
  • 5 Laboratory for Ocean Dynamics and Climate, Qingdao National Laboratory for Marine Science and Technology, Qingdao 266071, China
  • 6 College of Meteorology and Oceanography, National University of Defense Technology, Nanjing 210007, China
Published: 2018-06-25 doi: 10.1007/s13131-018-1217-1
Outline
收藏切换

Chinese Gaofen-3 (GF-3) is the first civilian satellite to carry C-band (5.3 GHz) synthetic aperture radar (SAR). During the period of August 2016 to December 2017, 1 523 GF-3 SAR images acquired in quad-polarization (vertical-vertical (VV), horizontal-horizontal (HH), vertical-horizontal (VH), and horizontal-vertical (HV)) mode were recorded, mostly around China’s seas. In our previous study, the root mean square error (RMSE) of significant wave height (SWH) was found to be around 0.58 m when compared with retrieval results from a few GF-3 SAR images in co-polarization (VV and HH) with moored measurements by using an empirical algorithm CSAR_WAVE. We collected a number of sub-scenes from these 1 523 images in the co-polarization channel, which were collocated with wind and SWH data from the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis field at a 0.125° grid. Through the collected dataset, an improved empirical wave retrieval algorithm for GF-3 SAR in co-polarization was tuned, herein denoted as CSAR_WAVE2. An additional 92 GF-3 SAR images were implemented in order to validate CSAR_WAVE2 against SWH from altimeter Jason-2, showing an about 0.52 m RMSE of SWH for co-polarization GF-3 SAR. Therefore, we conclude that the proposed empirical algorithm has a good performance for wave retrieval from GF-3 SAR images in co-polarization.

Gaofen-3  /  synthetic aperture radar  /  significant wave height
Yexin SHENG, Weizeng SHAO, Shuai ZHU, Jian SUN, Xinzhe YUAN, Shuiqing LI, Jian SHI, Juncheng ZUO. Validation of significant wave height retrieval from co-polarization Chinese Gaofen-3 SAR imagery using an improved algorithm[J]. Acta Oceanologica Sinica, 2018 , 37 (6) : 1 -10 . DOI: 10.1007/s13131-018-1217-1
Full Text
Less
1 Introduction
Less
It is well known that synthetic aperture radar (SAR) has the capability of wind and wave monitoring (Chapron et al., 2001) in large swath coverage with a fine spatial resolution, especially in extreme sea states (Li et al., 2002; Hwang and Fois, 2015; Li, 2015; Shao et al., 2017a). To date, SAR data is available at C-band (5.3 GHz) Canadian Radarsat-2 (R-2), and European Sentinel-1 (S-1); X-band (9.8 GHz) German TerraSAR-X/TanDEM-X, Italian Cosmo-SkyMed and Korean Kompsat-5; and L-band (1.2 GHz) Japanese ALOS-2 satellite. Gaofen-3 (GF-3) SAR at C-band was launched by the China Academy of Space Technology (CAST) in August 2016, and can operate in 12 imaging modes with a fine spatial resolution of image up to 1 m. It has a 755-km orbit height above the earth's surface with a 26-day repeat cycle. Recently, preliminary analysis of marine applications using GF-3 SAR data have been achieved, in particular, for wind (Wang et al., 2017; Ren et al., 2017; Shao et al., 2018) and wave monitoring (Yang et al., 2017; Shao et al., 2017b).
Based on a good understanding of the wave imaging mechanism on SAR, including tilt modulation (Lyzenga, 1986), hydrodynamic modulation (Feindt et al., 1986) and velocity bunching (Alpers et al., 1981; Alpers and Bruening, 1986), wave retrieval algorithms have been thoroughly studied over recent decades. Basic scattering physics is widely used in theoretical-based wave retrieval algorithms, e.g., Max-Planck Institute Algorithm (MPI) (Hasselmann and Hasselmann, 1991; Hasselmann et al., 1996), the semi parametric retrieval algorithm (SPRA) (Mastenbroek and De Valk, 2000), the parameterized first-guess spectrum method (PFSM) (Sun and Guan, 2006; Shao et al., 2015; Lin et al., 2017) and the partition rescaling and shift algorithm (PARSA) (Schulz-Stellenfleth et al., 2005; Li et al., 2010), which are independent of radar frequency and imaging polarization. However, velocity bunching is a non-linear modulation, that causes waves of a shorter than specific wavelength to be undetectable in the azimuth direction (or satellite flight direction) and a cutoff in the SAR intensity spectrum (Alpers and Bruening, 1986; Hasselmann and Hasselmann, 1991). The idea behind these theoretical-based algorithms is directly inverting the SAR intensity spectrum into the wave spectrum after employing a “first-guess” wave spectrum, which is considered to be the compensation for loss in the SAR intensity spectrum due to non-linear effect of velocity bunching. The algorithms MPI and PARSA take the simulation from a numeric wave model, while a prior wave spectrum is produced by using a parameterized empirical function in the schemes of algorithms SPRA and PFSM, such as the Jonswap spectrum (Hasselmann and Hasselmann, 1985). Therefore, they are limitedly applied in the operation system, because the quality of the “first-guess” wave spectrum determines the SAR-derived wave spectrum. Moreover the “first-guess” wave spectrum is not reliable in the presence of other marine phenomena. Ocean wave parameters, e.g., significant wave height (SWH) and mean wave period (MWP), are calculated from the SAR-derived wave spectrum.
An empirical wave retrieval algorithm for C-band ERS SAR is proposed by Schulz-Stellenfleth et al. (2007), denoted as CWAVE_ERS. In particular, CWAVE has been tuned for ENVISAT Advanced SAR (ASAR) (Li et al., 2011) and S-1 SAR (Stopa and Mouche, 2017). The CWAVE model is designed to be an empirical function, in which the sea state parameter SWH is connected with a set of variables, including normalized radar cross section (NRCS), variance of the normalized SAR image and several orthonormal functions derived from the two-dimensional SAR spectrum. The advantage is that SWH can be directly retrieved from SAR without calculating the modulation transfer function (MTF) of each SAR mapping modulation. However, CWAVEs have only been validated for SAR data acquired in wave mode until now. Following the idea of CWAVE, researchers have recently exploited the empirical algorithms XWAVE for X-band SAR (Bruck and Lehner, 2015; Pleskachevsky et al., 2016; Shao et al., 2017c).
Recent research has revealed that the azimuthal cutoff wavelength is derived to be proportional to the second moment of a wave spectrum (Hasselmann and Hasselmann, 1991; Marghany et al., 2002). On the other hand, SWH is calculated by integrating a wave spectrum according to traditional wave theory. Therefore, SWH is theoretically related to cutoff wavelength in the azimuth direction. Interestingly, several studies have made an attempt to retrieve SWH through azimuthal cutoff wavelength (Wang et al., 2012; Ren et al., 2015; Grieco et al., 2016; Stopa et al., 2016). The dependences of radar incidence angle and wave propagation direction on azimuthal cutoff wavelength were investigated in our previous study using theoretical analysis and simulation experiment (Shao et al., 2016). Then we constructed an empirical wave retrieval algorithm, denoted as CSAR_WAVE, which was tuned through VV-polarization S-1 SAR image and collocated measurements from the National Data Buoy Center (NDBC) buoys of the National Oceanic and Atmospheric Administration (NOAA). The preliminary assessment showed that CSAR_WAVE is applicable for GF-3 SAR with around 0.58 m root mean square error (RMSE) of the retrieved SWH compared with the NDBC buoy measurements of NOAA (Shao et al., 2017b). However, the accuracy of the retrieval results is expected to be further improved for the operational application of GF-3 SAR, as the SAR-derived product is dedicated to oceanography research, especially in coastal waters. Therefore, in this study, we have developed an improved wave retrieval algorithm for GF-3 SAR in co-polarization (VV and HH).
The remaining part of this paper is organized as follows: collected datasets are briefly described in Section 2. Section 3 shows the methodology of derivation of the empirical algorithm. In this section, the process of tuning the empirical algorithm for co-polarization GF-3 SAR is also presented. Then the validation of the retrieved SWHs using the proposed algorithm and other three existing empirical algorithms, are shown in Section 4. Conclusions are summarized in Section 5.
2 Description of dataset
Less
Since GF-3 SAR was launched in 2016 by CAST, during the period of August 2016 to December 2017 a number of images acquired in quad-polarization mode (QPS-I/II) (vertical-vertical (VV), horizontal-horizontal (HH), vertical-horizontal (VH), and horizontal-vertical (HV)) have been recorded. Most of these GF-3 SAR images were located around China’s seas and they were processed as Level-1A (L-1A) products, which have a standard pixel of 8 m and 25 m for QPS-I and QPS-II mode, respectively. Because the SAR backscattering signature from a sea surface in co-polarization is more sensitive than that in cross-polarization (VH and HV), the collected GF-3 SAR images in VV- and HH-polarization channel are used in our study. Equation 1 is used for calculating the NRCS of a co-polarization GF-3 SAR intensity image.
${\sigma ^{\rm{0}}}{\rm{ = }}D{N^{\rm{2}}}{\left({\frac{M}{{{\rm{32}}\;{\rm{767}}}}} \right)^{\rm{2}}} - N, $
where σ0 is the NRCS united in dB, DN is the SAR-measured intensity, M and N are the calibrated constants stored in the annotated file with the original SAR intensity image.
As an example, a quick-look image of the calibrated GF-3 SAR image acquired in QPS-I mode at 10:40 UTC on 18 January 2017 in VV- and HH-polarization is shown in Fig. 1a and Fig. 1b, respectively. It was found that wind direction is vertical to two-dimensional SAR image spectra for wavelengths between 800 m and 3 000 m at peaks (Alpers and Brümmer, 1994), indicating wind direction can be directly measured from SAR. However, the SAR-derived wind direction has a 180° ambiguity. The European Centre for Medium-Range Weather Forecasts (ECMWF) provides global reanalysis wind data with a fine spatial resolution of 0.125°×0.125° at intervals of six hours, which is employed to remove that ambiguity. The wind speed (U10) at 10 m-height above sea surface can be inverted by using the combination method proposed in our previous study (Shao et al., 2014), which is based on the geophysical model function (GMF) CMOD5 (Hersbach et al., 2007) and CMOD4 (Stoffelen and Anderson, 1997). The colored vectors shown in Fig. 1 represent the SAR-derived wind fields. Note that it is necessary to use the polarization ratio (PR) at C-band (Zhang et al., 2011) together with GMF to retrieve the wind field from an HH-polarization GF-3 SAR image.
In our study, all the GF-3 SAR images are divided into a number of sub-scenes with a spatial coverage of about 5 km×5 km. These extracted sub-scenes are collocated with 0.125° gridded ECMWF reanalysis wave data at intervals of six hours. It is necessary to ensure that the sub-scenes covering the locations of the ECMWF reanalysis grids data are calculated by bilinear interpolation in temporal scale, as there is a time difference between the GF-3 SAR imaging time and the interval time of the ECMWF reanalysis grids data. Then we have more than ten thousand matchups, which are treated as a dataset for tuning an improved algorithm for wave retrieval from GF-3 SAR images. Figure 2 shows the ECMWF reanalysis wind and wave map at 06:00 UTC on 18 January 2017, in which the black rectangle represents the spatial coverage of a GF-3 SAR image located in the South China Sea as exhibited in Fig.1. It should be noted that the SWH from the ECMWF reanalysis data goes up to 4 m, therefore, GF-3 SAR images at low to moderate sea states are included in the dataset. Recently, a new approach for SWH retrieval in hurricanes has been constructed through studying the relationship between SWH and NRCS (Romeiser et al., 2015). As mentioned by the authors, this is still to be improved due to the complicated non-linear effect of waves at extreme sea states.
The high-precision ocean altimetry on Jason-2 launched in 2008 is a marine observation system over global sea, which is a follow-on satellite of the oceanography monitoring mission of Jason-1. So far, Operational Geophysical Data Record (OGDR) derived from the Jason-2 satellite track is a near real-time operational product, in particular including more reliable SWH data which is better than that of Jason-1 by about 7% (Abdalla et al., 2010). This high-quality product is essentially dedicated to oceanography research. An additional 91 quad-polarization GF-3 SAR images were collected and these GF-3 SAR images cover the footprints of altimeter Jason-2, which were implemented in order to validate the improved algorithm in our study.
3 Methodology
Less
In this section, the methodology of derivation of an improved algorithm is presented, which is based on two existing empirical wave retrieval algorithms CWAVE and CSAR_WAVE. Then the improved algorithm, denoted CSAR_WAVE2, is tuned for co-polarization GF-3 SAR.
3.1 Algorithm CWAVE
As mentioned in Section 1, algorithms MPI, SPRA, PFSM and PARSA rely on prior information on a wave spectrum, e.g., numeric simulation from a wave model and computation from a parametric wave function. In the operational application, they take some time to produce a “first-guess” wave spectrum and on the non-linear inversion of an SAR spectrum into a wave spectrum (Hasselmann and Hasselmann, 1991; Hasselmann et al., 1996). Moreover, it is difficult to improve the accuracy of SWH retrieval in the physics aspect of theoretical-based algorithms. In practice, empirical models are routine operations for marine applications of Scatterometer and SAR, such as GMFs for wind retrieval (Stoffelen and Anderson, 1997; Hersbach et al., 2007). The GMF CWAVE family, e.g., CWAVE_ERS (Schulz-Stellenfleth et al., 2007) for ERS SAR and CWAVE_ENV (Li et al., 2011) for ENVISAT-ASAR, were originally exploited by the SAR oceanography group at the German Aerospace Center (DLR), which allows for direct retrieval of wave parameters from SAR wave mode data without calculating the complex MTF of each SAR mapping modulation.
In a SAR image, sea state measurement S can be determined by a set of imaging parameters si (s1, s2, …, sn) with a coefficient vector ai (a0, a1, …, an). Due to the modulation of velocity bunching, non-linearity among different imaging parameters is also included by adding the products of different imaging parameters si with a coefficient vector ai, j (i≤j≤n). Based on this assumption, the function of CWAVE principally follows the multiple-regression method stated as
$S{\rm{ = }}{a_{\rm{0}}}{\rm{ + }}\mathop \sum \limits_{i{\rm{ = 1}}}^n {a_i}{\rm{ \times }}{s_i}{\rm{ + }}\mathop \sum \limits_{i, \,\, j{\rm{ = 1}}}^n {a_{i, \,\, j}}{\rm{ \times }}{s_i}{\rm{ \times }}{s_j}. $
In the CWAVE models, imaging parameters si include NRCS σ0, and variance of the normalized SAR image cvar, both of which directly contribute to sea state, and a set of orthonormal functions derived from the two-dimensional SAR spectrum. cvar is defined as follows:
$cvar{\rm{ = var}}\left({\frac{{I - \bar I}}{{\bar I}}} \right), $
where I is the pixel intensity of a SAR image and ${\bar I}$ is the average of I. The coefficients in CWAVE models were tuned for ERS and ENVISAT-ASAR wave mode data acquired in VV-polarization at a fixed incidence angle of 23°. Therefore, CWAVE needs to be retuned for other SAR data at various incidence angles, such as CWAVE_S1 for S-1 SAR (Stopa and Mouche, 2017).
3.2 Algorithm CSAR_WAVE
The relationship between cutoff wavelength in azimuth direction λc and SWH was demonstrated in the study proposed by Hasselmann and Hasselmann (1991):
${\text{λ} _{\text{c}}} = \text{π} \text{β} \sqrt {\int {{{\left| {T_\text{ω}^v} \right|}^{\text{2}}}{S_\text{ω} }{\text{d}}\text{ω} } } ,$
where β is the satellite range-to-velocity parameter, |Tωv| is the velocity bunching transfer function, ω is wave frequency and Sω is the one-dimensional wave spectrum. In the imaging process, λc can be estimated by fitting a one-dimensional SAR spectrum with a Gaussian fit function (Sun and Kawamura, 2009). The Gaussian fit function has the formulation exp{π(kx/kc)}, in which kx is the azimuthal wavenumber and kc=2π/λc is the azimuthal cutoff wavenumber. Through analyzing a number of recorded ENVISAT-ASAR wave mode data, recent research has revealed that λc provides meaningful information about the sea state, even at large sea states (>250 m) (Stopa et al., 2016).
SWH can be calculated by integrating wave spectrum Sω,
$ SWH{\text{ = 4}}\sqrt {{\text{ }}\int {{S_\text{ω} }} {\text{d}}\text{ω} } .$
Theoretically, SWH is related to λc through the above two equations. Recently, several algorithms have been developed by using the λc to estimate SWH for ENVISAT-ASAR (Wang et al., 2012), quad-polarization R-2 SAR (Ren et al., 2015) and S-1 SAR (Grieco et al., 2016; Stopa and Mouche, 2017).
The dependency of λc, radar incidence angle θ and peak wave direction relative to range direction $\varphi $ on SWH was simulated through the widely used Jonswap wave spectrum model (Hasselmann and Hasselmann, 1985). It was found that SWH is linearly related with λc/β, while SWH has a positive and negative relationship with θ and φ respectively (Shao et al., 2016). The semi-empirical wave retrieval algorithm, denoted as CSAR_WAVE, was originally developed for S-1 SAR in our previous study. The formulation of CSAR_WAVE is designed as a first-order linear function,
$SWH = \left({\frac{{{\text{λ} _{\rm c}}}}{\text{β} }} \right)\left({{A_1} + {A_2}\sin \text{θ} + {A_3}\cos 2\text{φ} } \right) + {A_4}, $
where coefficients A are determined from S-1 SAR image collocated NDBC buoys of NOAA. It was reported by Shao et al. (2017b) that the RSME of SWH is 0.58 m and 0.57 m when using CSAR_WAVE for GF-3 SAR in VV- and HH-polarization respectively, as the retrieved SWHs are validated against the NDBC buoys of NOAA around U.S. waters.
3.3 Tuning the improved algorithm
In order to enhance the sensitivity of non-linearity on SWH in an empirical algorithm, the formulation of a CWAVE model is basically employed. However, imaging parameters si in the CWAVE model are set as a vector (U10, σ0, cvar, λc/β, sinθ, cos2$\varphi $, λSAR) for practical application, in which three factors, e.g., U10, σ0 (united in dB) and cvar, are directly related with sea state (Li et al., 2010; Grieco et al., 2016; Stopa et al., 2016). Besides, the dependences of other factors on SWH, including λc/β, θ and $\varphi $, have been already investigated by Shao et al. (2016). In particular, λSAR represents the SAR length at peaks of the SAR spectrum, which is also assumed to be an essential factor in SWH, according to the derivation model through the SAR imaging mechanism of ocean wave, as referred to in Eq. (16) proposed by Wang et al. (2012).
In total, we have obtained more than ten thousand sub-scenes extracted from GF-3 images in co-polarization channel with collocated ECMWF reanalysis SWH data. During the process, three variables, i.e., λc, $\varphi $ and λSAR, were derived from the SAR intensity spectrum. The sub-scene extracted from the case exhibited in Fig. 1, which is acquired in VV-polarization, is shown in Fig. 3a. The corresponding two-dimensional SAR spectrum of the sub-scene is shown in Fig. 3b, in which $\varphi $ and λSAR can be directly obtained. The Gaussian fitted result of λc is illustrated in Fig. 3c.
The matchup dataset is used to determine the 36 coefficients ai, j (i≤j≤7) in Eq. (2) by using the least-squares method, in which subscripts (1, 2, …, 7) represent the corresponding variables (U10, σ0, cvar, λc/β, sinθ, cos2$\varphi $, λSAR), e.g., a12 is the coefficient for the term of U10×σ0. The tuned results in the improved algorithm, denoted as CSAR_WAVE2, are shown in Table 1 for co-polarization GF-3 SAR.
Figure 4 shows the fitting results of CSAR_WAVE2 compared with ECMWF reanalysis SWH in our data collection. It is found that the correlation (COR) between the ECMWF reanalysis data and the simulated values is around 0.72 for co-polarization GF-3 SAR. Under these circumstances, we think the improved algorithm CSAR_WAVE2 is suitable for SWH retrieval from co-polarization GF-3 SAR images.
4 Validation
Less
With reference to the application process of existing CWAVE and CSAR_WAVE models, the process of SWH retrieval by our use of CSAR_WAVE2 is roughly illustrated in Fig. 5. We first show the quick-look image of the VV-polarization GF-3 SAR image acquired at 20:49 UTC on 26 July 2017 in Fig. 6a. The inverted wave map for this case when using CSAR_WAVE2 is shown in Fig. 6b, in which the several small colored rectangles represent the SWH data measured from the altimeter Jason-2 footprints. It is found that the SAR-derived SWH is close to the SWH data of Jason-2. In particular, the trend of the inverted wave map is consistent with that following the track of the Jason-2 footprints.
In addition, we have applied CSAR_WAVE2 to a total of 91 available GF-3 SAR images and compared the results with those from the SWH data from altimeter Jason-2. In Fig. 7, the RMSE of SWH is 0.51 m for VV-polarization and the RMSE of SWH is 0.52 m for HH-polarization. The reported accuracy of SWH for C-band SAR is an RMSE of SWH of 0.55 m as validated against measurements from moored buoys using the PFSM algorithm (Lin et al., 2017) and RMSE is 0.51 m when comparing the wave retrievals with the WAM model predictions (Schulz-Stellenfleth et al., 2005) using the PARSA algorithm. It is indicated that CSAR_WAVE2 has a better accuracy of SWH retrieval than that using theoretical-based algorithms. In particular, it is applicable without calculating the complex MTF of each mapping modulation.
We also compared the SAR-derived results with SWH from Jason-2 by using the existing three empirical algorithms proposed by Wang et al. (2012), Ren et al. (2015) and Grieco et al. (2016). All of these algorithms were developed based on azimuthal cutoff wavelength and tuned through R-2 and S-1 SAR data acquired in VV-polarization. Figure 8 shows that the RMSE of SWH is 0.70 m, 0.62 m and 0.61 m using the algorithms by Wang et al. (2012), Ren et al. (2015) and Grieco et al. (2016), respectively. And a comparison between SAR-derived SWHs and measurements from NDBC buoys of NOAA shows an approximate 0.58 m RMSE of SWH for co-polarization using CSAR_WAVE (Shao et al., 2017b). This analysis shows that these algorithms all perform less well than the results achieved using CSAR_WAVE2, when non-linear higher-order corrections on sea state are included in CSAR_WAVE2. Therefore, it is recommended that CSAR_WAVE2 is applied operationally for wave retrieval from GF-3 SAR images in co-polarization. However, it is necessary to establish that there are no available data at high sea states in the fitting and validation procedure. CSAR_WAVE2 is expected to be further adopted for high sea states as the non-linearity is higher than at low and moderate sea states, especially in typhoons and hurricanes.
5 Summary and conclusion
Less
In the preliminary assessment (Shao et al., 2017b), the RMSE of SWH was around 0.58 m for GF-3 SAR when using the empirical wave retrieval algorithm CSAR_WAVE as validated against buoy measurements, which was tuned for S-1 SAR in VV-polarization. As for the operational application of GF-3 SAR, it is essential to reduce the retrieval error of the SWH for oceanic and coastal monitoring.
In this study, 1 523 GF-3 SAR images acquired in quad-polarization mode were collected during the period of August 2016 to December 2017. More than ten thousand sub-scenes from these images in the co-polarization channel were collocated with SWH from ECMWF reanalysis data at a 0.125° grid with SWH up to 4 m. Through the dataset, an improved wave retrieval algorithm for GF-3 SAR, denoted as CSAR_WAVE2, was developed. Seven variables, which are explicitly related to sea state and can be directly obtained from a SAR image, were selected for the CSAR_WAVE2 model. CSAR_WAVE2 is more than an updated version of CSAR_WAVE, as the formulation of function has been rigorously redesigned and non-linear higher-order corrections on sea state have been implemented. The COR is 0.72 and 0.71 for VV- and HH-polarization respectively, when the simulated SWH using CSAR_WAVE2 is compared with ECMWF reanalysis SWH data, indicating that CSAR_WAVE2 can be applied for wave retrieval from GF-3 SAR image in co-polarization.
An additional 92 GF-3 SAR images were collected, which cover the footprint of the altimeter Jason-2 mission. Validation shows that the RMSE of SWH is 0.51 m and 0.52 m for GF-3 SAR in VV- and HH-polarization respectively. We also compared the retrieval results with SWH of Jason-2 using three existing empirical algorithms (Wang et al., 2012; Ren et al., 2015; Grieco et al., 2016), showing a 0.60–0.70 m RMSE of SWH. As a result, it is concluded that the accuracy of retrieved SWH from co-polarization GF-3 SAR has been significantly improved using CSAR_WAVE2 at low to moderate sea states.
The applicability of CSAR_WAVE2 will be further investigated for various GF-3 SAR data, e.g., Spotlight Mode (SL), Standard Stripmap (SS), Wide Scan (WSC), Global Observing Mode (GLO) and Wave Mode (WAV). Recently, GF-3 SAR has captured several typhoons by the National Ocean Satellite Application Center (NSOAS) around China’s seas. Therefore, the applicability of CSAR_WAVE2 will be further investigated and can be adopted for high sea states in the near future.
Acknowledgements
Less
Gaofen-3 synthetic aperture radar (SAR) images are collected through an authorized account issued by the National Ocean Satellite Application Center (NSOAS) under the contract of Specific Project of Chinese High Resolution Earth Observation System (No. 41-Y20A14-9001-15/16) via http://dds.nsoas.org.cn. The authers greatly appreciate the European Centre for Medium-Range Weather Forecasts (ECMWF) for providing reanalysis wind and wave data at a 0.125 grid, which were openly downloaded via http://www.ecmwf.int. Operational Geophysical Data Record (OGDR) wave data from Jason-2 mission were accessed via https://data.nodc.noaa.gov. The authers thank Cui Limin (NSOAS) and Li Huan (National Marine Data and Information Service) for the helpful discussions.
Funding
Less
  • The National Key Research and Development Program of China under contract Nos 2016YFC1401905 and 2017YFA0604901; the National Natural Science Foundation of China under contract Nos 41776183, 41676014, 41606024 and 41506033; the National Social Science Foundation of China under contract No. 15ZDB170.
References
Less
Abdalla S, Janssen P A E M, Bidlot J R. 2010. Jason-2 OGDR wind and wave products: monitoring, validation and assimilation. Marine Geodesy, 33(S1): 239–255
Alpers W, Brümmer B. 1994. Atmospheric boundary layer rolls observed by the synthetic aperture radar aboard the ERS-1 satellite. Journal of Geophysical Research, 99(C6): 12613–12621
Alpers W R, Ross D B, Rufenach C L. 1981. On the detectability of ocean surface waves by real and synthetic aperture radar. Journal of Geophysical Research: Oceans, 86(C7): 6481–6498
Alpers W R, Bruening C. 1986. On the relative importance of motion-related contributions to the Sar imaging mechanism of ocean surface waves. IEEE Transactions on Geoscience and Remote Sensing, GE-24(6): 873–885
Bruck M, Lehner S. 2015. TerraSAR-X/TanDEM-X sea state measurements using the XWAVE algorithm. International Journal of Remote Sensing, 36(15): 3890–3912
Chapron B, Johnsen H, Garello R. 2001. Wave and wind retrieval from SAR images of the ocean. Annales Des Télécommunications, 56(11–12): 682–699
Feindt F, Schröter J, Alpers W. 1986. Measurement of the ocean wave-radar modulation transfer function at 35 GHz from a sea-based platform in the North Sea. Journal of Geophysical Research: Oceans, 91(C8): 9701–9708
Grieco G, Lin W, Migliaccio M, et al. 2016. Dependency of the sentinel-1 azimuth wavelength Cut-off on significant wave height and wind speed. International Journal of Remote Sensing, 37(21): 5086–5104
Hasselmann K, Hasselmann S. 1991. On the nonlinear mapping of an ocean wave spectrum into a synthetic aperture radar image spectrum and its inversion. Journal of Geophysical Research: Oceans, 96(C6): 10713–10729
Hasselmann S, Hasselmann K. 1985. Computations and parametrizations of the nonlinear energy transfer in a gravity wave spectrum: Part I. a new method for efficient computations of the exact nonlinear transfer integral. Journal of Physical Oceanography, 15: 1369–1377
Hasselmann S, Bruning C, Hasselmann K. 1996. An improved algorithm for the retrieval of ocean wave spectra from synthetic aperture radar image spectra. Journal of Geophysical Research: Oceans, 101(C7): 16615–16629
Hersbach H, Stoffelen A, De Haan S. 2007. An improved C-band scatterometer ocean geophysical model function: CMOD5. Journal of Geophysical Research: Oceans, 112(C3): C03006
Hwang P A, Fois F. 2015. Surface roughness and breaking wave properties retrieved from polarimetric microwave radar backscattering. Journal of Geophysical Research: Oceans, 120(5): 3640–3657
Li Xiaofeng, Pichel W, He Mingxia, et al. 2002. Observation of hurricane-generated ocean swell refraction at the gulf stream north wall with the RADARSAT-1 synthetic aperture radar. IEEE Transactions on Geoscience and Remote Sensing, 40(10): 2131–2142
Li Xiaoming, Koenig T, Schulz-Stellenfleth J, et al. 2010. Validation and intercomparison of ocean wave spectra inversion schemes using ASAR wave mode data. International Journal of Remote Sensing, 31(17): 4969–4993
Li Xiaoming, Lehner S, Bruns T. 2011. Ocean wave integral parameter measurements using envisat ASAR wave mode data. IEEE Transactions on Geoscience and Remote Sensing, 49(1): 155–174
Li Xiaofeng. 2015. The first sentinel-1 SAR image of a typhoon. Acta Oceanologica Sinica, 34(1): 1–2
Lin Bo, Shao Weizeng, Li Xiaofeng, et al. 2017. Development and validation of an ocean wave retrieval algorithm for VV-polarization sentinel-1 SAR Data. Acta Oceanologica Sinica, 36(7): 95–101
Lyzenga D R. 1986. Numerical simulation of synthetic aperture radar image spectra for ocean waves. IEEE Transactions on Geoscience and Remote Sensing, GE-24(6): 863–872
Marghany M, Ibrahim Z, Van Genderen J. 2002. Azimuth cut-off model for significant wave height investigation along coastal water of Kuala Terengganu, Malaysia. International Journal of Applied Earth Observation and Geoinformation, 4(2): 147–160
Mastenbroek C, De Valk C F. 2000. A semiparametric algorithm to retrieve ocean wave spectra from synthetic aperture radar. Journal of Geophysical Research: Oceans, 105(C2): 3497–3516
Pleskachevsky A L, Rosenthal W, Lehner S. 2016. Meteo-marine parameters for highly variable environment in coastal regions from satellite radar images. ISPRS Journal of Photogrammetry and Remote Sensing, 119(2): 464–484
Ren Lin, Yang Jingsong, Zheng Gang, et al. 2015. Significant wave height estimation using azimuth Cutoff of C-band RADARSAT-2 single-polarization SAR images. Acta Oceanologica Sinica, 34(12): 93–101
Ren Lin, Yang Jingsong, Mouche A, et al. 2017. Preliminary analysis of Chinese GF-3 SAR Quad-polarization measurements to extract winds in each polarization. Remote Sensing, 9(12): 1215
Romeiser R, Graber H C, Caruso M J, et al. 2015. A new approach to ocean wave parameter estimates from C-band ScanSAR images. IEEE Transactions on Geoscience and Remote Sensing, 53(3): 1320–1345
Schulz-Stellenfleth J, Lehner S, Hoja D. 2005. A parametric scheme for the retrieval of two-dimensional ocean wave spectra from synthetic aperture radar look cross spectra. Journal of Geophysical Research: Oceans, 101(C5): C05004
Schulz-Stellenfleth J, König T, Lehner S. 2007. An empirical approach for the retrieval of integral ocean wave parameters from synthetic aperture radar data. Journal of Geophysical Research: Oceans, 112(C3): C03019
Shao Weizeng, Li Xiaofeng, Hwang P, et al. 2017a. Bridging the gap between cyclone wind and wave by C-band SAR measurements. Journal of Geophysical Research: Oceans, 122(8): 6714–6724
Shao Weizeng, Li Xiaofeng, Sun Jian. 2015. Ocean wave parameters retrieval from TerraSAR-X images validated against buoy measurements and model results. Remote Sensing, 7(10): 12815–12828
Shao Weizeng, Sheng Yexin, Sun Jian. 2017b. Preliminary assessment of wind and wave retrieval from Chinese Gaofen-3 SAR imagery. Sensors, 17(8): 1705
Shao Weizeng, Sun Jian, Guan Changlong, et al. 2014. A method for sea surface wind field retrieval from SAR image mode data. Journal of Ocean University of China, 13(2): 198–204
Shao Weizeng, Wang Jing, Li Xiaofeng, et al. 2017c. An empirical algorithm for wave retrieval from Co-polarization X-Band SAR imagery. Remote Sensing, 9(7): 711
Shao Weizeng, Yuan Xinzhe, Sheng Yexin, et al. 2018. Development of wind speed retrieval from cross-polarization Chinese Gaofen-3 synthetic aperture radar in typhoons. Sensors, 18(2): 412
Shao Weizeng, Zhang Zheng, Li Xiaofeng, et al. 2016. Ocean wave parameters retrieval from sentinel-1 SAR imagery. Remote Sensing, 8(9): 707
Stoffelen A, Anderson D. 1997. Scatterometer data interpretation: estimation and validation of the transfer function CMOD4. Journal of Geophysical Research: Oceans, 102(C3): 5767–5780
Stopa J E, Ardhuin F, Chapron B, et al. 2016. Estimating wave orbital velocity through the azimuth cutoff from space-borne satellites. Journal of Geophysical Research: Oceans, 120(11): 7616–7634
Stopa J E, Mouche A. 2017. Significant wave heights from sentinel-1 SAR: validation and applications. Journal of Geophysical Research: Oceans, 122(3): 1827–1848
Sun Jian, Guan Changlong. 2006. Parameterized first-guess spectrum method for retrieving directional spectrum of swell-dominated waves and huge waves from SAR images. Chinese Journal of Oceanology and Limnology, 24(1): 12–20
Sun Jian, Kawamura H. 2009. Retrieval of surface wave parameters from SAR images and their validation in the coastal seas around Japan. Journal of Oceanography, 65(4): 567
Wang He, Yang Jingsong, Mouche A, et al. 2017. GF-3 SAR ocean wind retrieval: the first view and preliminary assessment. Remote Sensing, 9(7): 694
Wang He, Zhu Jianhua, Yang Jingsong, et al. 2012. A semiempirical algorithm for SAR wave height retrieval and its validation using envisat ASAR wave mode data. Acta Oceanologica Sinica, 31(3): 59–66
Yang Jingsong, Wang Juan, Ren Lin. 2017. The first quantitative remote sensing of ocean internal waves by Chinese GF-3 SAR satellite. Acta Oceanologica Sinica, 36(1): 118
Zhang Biao, Perrie W, He Yijun. 2011. Wind speed retrieval from RADARSAT-2 quad-polarization images using a new polarization ratio model. Journal of Geophysical Research: Oceans, 116(C8): C08008
Appendix
Less
Year 2018 volume 37 Issue 6
PDF
48
25
Cite this Article
BibTeX
Article Info
doi: 10.1007/s13131-018-1217-1
  • Receive Date：2018-01-23
  • Online Date：2026-04-14
  • Published：2018-06-25
Article Data
Affiliations
History
  • Received：2018-01-23
  • Accepted：2018-03-21
Funding
The National Key Research and Development Program of China under contract Nos 2016YFC1401905 and 2017YFA0604901; the National Natural Science Foundation of China under contract Nos 41776183, 41676014, 41606024 and 41506033; the National Social Science Foundation of China under contract No. 15ZDB170.
Affiliations
    1 Marine Science and Technology College, Zhejiang Ocean University, Zhoushan 316000, China
    2 Physical Oceanography Laboratory/CIMST, Ocean University of China and Qingdao National Laboratory for Marine Science and Technology, Qingdao 266100, China
    3 National Satellite Ocean Application Service, State Oceanic Administration, Beijing 100081, China
    4 Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071, China
    5 Laboratory for Ocean Dynamics and Climate, Qingdao National Laboratory for Marine Science and Technology, Qingdao 266071, China
    6 College of Meteorology and Oceanography, National University of Defense Technology, Nanjing 210007, China

Corresponding:

References
Share
https://castjournals.cast.org.cn/joweb/aos/EN/10.1007/s13131-018-1217-1
Share to
QR

Scan QR to access full text

Cite this article
BibTeX
Citations
表12种不同金属材料的力学参数
科
Family		 属数
Number of
genus		 种数
Number of
species		 占总种数比例
Percentage of
total species (%)		 属
Genus		 种数
Number of
species		 占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏
  • BibTeX
  • EndNote
  • RefWorks
  • TxT
Links
Articles
Latest Articles
Most Read
Collections
Updates
Events
News
Multimedia
About
About Us
Contact
No. 86 Xueyuan South Road, Haidian District, Beijing
100081
010-62199257
qkjq@cast.org.cn

Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House