Sea Surface Wind Field Retrieval and Offshore Wind Energy Resource Assessment Based on C-band SAR Data

Author:Zhang Kangyu

Supervisor:Huang Jingfeng


Degree Year:2019





Spaceborne Synthetic Aperture Radar(SAR)has the ability to provide high spatial resolution sea surface wind fields.However,there are still the following problems in C-band SAR wind speed retrieval research:1)a consensus has not been reached on the optimal wind speed retrieval model for co-polarized SAR data;2)the existing cross-polarization wind speed retrieval models are not applicable to wind speed retrieval from Sentinel-1 cross-polarized SAR images;3)the spatial scale effect on SAR wind speed retrieval has rarely been studied.Thus,this study focuses on investigating the above issues.The coastal areas of Zhejiang Province and Fujian Province in China were used as test sites to evaluate the feasibility of using SAR images for offshore wind energy resources assessment.The main research contents and conclusions of this paper are as follows:(1)The optimal wind speed retrieval model for Sentinel-1 co-polarized SAR data was proposed.This paper compared the performances of four commonly used C-band model(CMOD)functions,which are CMOD4,CMOD-IFR2,CMOD5 and CMOD5.N,based on 710 scenes of Vertical Transmitting and Vertical Receiving(VV)polarized SAR images of the Sentinel-1 Interferometric Wide-swath(IW)mode;and compared the performances of 32 kinds CMOD + polarization ratio model combinations and one wind direction-independent HH polarization wind speed retrieval model based on 76 scenes of Sentinel-1 IW mode Horizontal Transmitting and Horizontal Receiving(HH)polarization SAR images.The results show that CMOD5.N is the optimal model for wind speed retrieval from Sentinel-1 VV polarized SAR and the root mean square error(RMSE)of the retrieved wind speeds is 1.40m/s in the wind speed range of 0~25m/s.For Sentinel-1 HH polarized SAR,the combination of CMOD4 and the polarization ratio model proposed by Mouche et 2005,which considers the effect of wind direction and incident angle,is the best HH polarization wind speed retrieval method,by which the RMSE of retrieved speeds is 1.45m/s.(2)A sub-swath based wind speed retrieval model for Sentinel-1 VH-polarized SAR data was developed.Due to the high noise equivalent sigma zero(NESZ)of the IW mode images and the different average NESZ of the three subs-waths in IW mode images,the existing cross-polarized wind speed retrieval model is not applicable for the retrieval of wind speeds from Sentinel-1 IW mode cross-polarized images.This paper evaluated the capability of Sentinel-1 VH-polarized SAR signals,involving 728 scenes in the IW mode,for ocean surface wind speed retrieval using a novel Sub-swath-based C-band Cross-polarized Ocean model(S-C2PO).When compared with in situ measurements,it is observed that wind speed retrieval accuracy varies progressively along swath,with the most accurate wind speed retrievals being derived from Sub-swath 3(RMSE of 1.82 m/s),followed by Sub-swath 2(RMSE of 1.92 m/s),while Sub-swath 1 showed the lowest retrieval accuracy(RMSE of 2.37 m/s).The average RMSE of wind speeds retrieved from all the three sub-swaths is 2.08 m/s under low-to-high wind speed regimes(wind speeds<25 m/s).We further observed that the dependence of VH-polarized Normalized Radar Cross Section(NRCS)on incidence angle is attributable to the high and changing NESZ with incidence angle under low-to-moderate wind speed regimes.Strong VH-polarized radar signals could overcome the NESZ effect,thereby eliminating the dependence of VH-polarized NRCS on incidence angle under strong wind conditions.For Sentinel-1 IW mode VH-polarized data,the effect of NESZ could be ignored when wind speeds are greater than 15 m/s,as a better wind speed retrieval performance of these data has been recorded in this study at wind speeds greater than 10 m/s,owing to an RMSE below 1.6 m/s and biases ranging from-0.5 m/s to 0.5 m/s.(3)The spatial scale effect on wind speed retrieval from co-polarized SAR data was investigated.High spatial resolution wind fields derived from SAR instruments are crucial to a wide range of applications.However,the spatial scale effect on wind speed retrieval from SAR data has seldom been reported.For the purpose of understanding this issue,this study maked a quality assessment of wind speed retrieval accuracy at spatial resolutions ranging from 100 m to 50 km using Sentinel-1 IW mode co-polarized images.It is observed that the variance of wind speeds retrieved from co-polarized SAR images decrease rapidly from 100 m to 1 km,and thereafter levels off.With the spatial resolution decrease from 100 m to 1km,the RMSE of wind speeds retrieved from VV-polarized SAR images drop 19.4%and the RMSE of wind speeds retrieved from HH-polarized SAR images drop 42.4%.(4)The spatial scale effect on wind speed retrieval from cross-polarized SAR data is investigated.This paper presented a comparison strategy for investigating the influence of spatial resolutions on sea surface wind speed retrieval accuracy with cross-polarized SAR images.Firstly,the most suitable C-band cross-polarized ocean(C-2PO)model was selected between two C-2POs for the VH-polarized image dataset.Then,the VH-wind speeds retrieved by the selected C-2PO were compared with the VV-polarized sea surface wind speeds retrieved using the optimal CMOD5.N,which served as reference,at different spatial resolutions.Results show that the VH-polarized wind speed retrieval accuracy increases rapidly with the decrease in spatial resolutions from 100 m to 1 km,with a drop in RMSE of 41.5%.However,the improvement in wind speed retrieval accuracy levels off with spatial resolutions decreasing from 1 km to 5 km.Combining the results of the spatial scale effect on wind speed retrieval from co-polarized SAR data,it is demonstrated that the pixel spacing of 1 km may be the compromising choice for the tradeoff between the spatial resolution and wind speed retrieval accuracy for wind speed retrieval from SAR data.(5)The feasibility of using SAR images to investigate offshore wind resources in China was evaluated.A wind power density distribution map was obtained using a total of 2749 SAR images of Sentinel-1 A,Sentinel-1B and ENVISAT ASAR.Taking the wind power parameters calculated by five buoys in China’s offshore water as references,the relative error of SAR-derived average wind speeds and Weibull scale factors(A)are both within 5%,and the relative error of Weibull shape factors(k)and wind power densities(E)are within 6%and within 10%,respectively.