Offshore Sea Surface Wind Field Retrieval Study Based On Sentinel-1

Sea surface wind field is an important part of marine microwave remote sensing research.It is also one of the most significant physical parameters and basic elements in the study of ocean phenomena and factors such as sea flow field,salinity,sea-air coupling system,tropical hurricanes,frontal surfaces and eddies etc.In addition,in many marine applications,such as ocean mapping,meteorological forecasting and oil spill monitoring,sea surface wind field provides the initial parameter for numerical simulation and prediction models of various ocean dynamic parameters.Besides,the effect of offshore sea surface wind on human activities is critical,especially nearshore fisheries,port engineering works and maritime navigation are quite vulnerable during some extreme weather conditions such as typhoons and storm surges.Therefore,the retrieval of much more accurate wind field is particularly important.Many methods are used to acquire sea surface wind field recently.Compared with the data obtained by the buoys with sporadic coverage and scatterometers with low-resolution,synthetic aperture radar(SAR)has a very broad prospect because of its high resolution,strong penetrating ability and short revisiting period.Nowadays,SAR is becoming an effective technical means to explore sea surface wind and wave in small and medium scales.Sentinel-1 SAR launched by ESA is selected to extract the neutral wind field in the coastal area of the Yangtze River estuary and the typical high wind field in the northwest Pacific area respectively.Based on the mechanism of SAR remote sensing,this paper discusses the reliability and applicability of combining multi-level wavelet transform and Fourier transform,according to the results comparing SAR-retrieved sea surface wind field with ECMWF wind field and five sets of in-situ data.Three kinds of GMF models are used to retrieve the wind speed information,in order to explore the optimum CMOD model for the Yangtze River estuary offshore area for extracting long-term wind field data in the future.Additionally,it discusses the application of the specific SAR-retrieved high-wind field containing typhoon structure in the northwest Pacific area,and also put forward opinions and problems about the retrieval effect and saturation of NRCS.The research results included:(1)In the retrieval of neutral wind field,the overall SAR-retrieved wind field is better than the ECMWF wind field except for some SAR sub-images with complex texture features.The results of wind direction after multi-level wavelet transform are better with the RMSE at 31.6°,29.7° and 23.5° respectively.Among three levels of multi-level wavelet transform,the result including wind direction information after second level of WT and wind speed obtained by the CMOD series model is the best,with the RMSE at 0.8m/s.It also indicates that both CMOD-IFR2 and CMOD4 are suitable for the Yangtze River estuary offshore area with the RMSE at 1.08m/s and 1.05m/s,respectively.(2)In the retrieval of high wind field,the wind direction which is consistent with wind-induced streaks is whirled around the typhoon eye in a low-wind speed.What's more,the position of the typhoon eye captured by SAR-retrieved wind field is much more accurate than which of ECMWF wind field.Besides,the wind speed at the center of typhoon is consistent with that of the ECMWF wind field.However,outside of the typhoon eye,the wind field results are quite different from ECMWF wind field.Because the wind vector changed so fast during the typhoon.ECMWF wind field was proved that it delayed about 2.5h.SAR-retrieved wind speed is generally around 20m/s,rather than the ECMWF wind field which has strong coherence and increases to approximately 20m/s from the typhoon eye to the outside.(3)Deficiencies.of SAR wind field retrieval:For neutral wind field,the process of extracting wind direction relies on wind-induced streaks.When SAR images lack wind-induced streaks or some linear streaks with same scales are not caused by wind,SAR-induced wind direction probably contain error,which will affect wind speed inversion in the further step.We are supposed to reduce the need of wind-induced streaks and then use some alternative parameters such as Doppler frequency shift.For high wind field,wind-induced streaks are so clear that it is good for the process of retrieval.But co-polarized(HH or VV)SAR data influenced by the saturation problem of NRCS result in inaccurate retrieval results under a high wind speed.In addition,it is difficult to obtain the validation data to evaluate SAR-induced high-speed wind field retrieval results.