| It is of great practical and scientific significance to obtain accurate oceanic wind field.Satellite-borne SAR with high spatial resolution has a strong ability to mapping the surface wind field at fine scale,and its rich polarization information makes it possible to obtain the surface wind speed under various sea conditions.In this paper,high-resolution C-band SAR data are used to refine or develop geophysical model functions under different polarization modes,and the retrieved wind speeds are compared with buoy observations.The main contents and conclusions are as follows:(1)In order to improve the accuracy of SAR derived coastal wind speed,1696 RS-2 and 899 Sl-1A images are matched with in situ buoy measurements to refine the C-band VVpolarized geophysical model function,to make it more suitable for offshore surface wind speed retrieval.The statistical comparison between retrieved wind speed using C-SARMOD2 and buoy observation shows a small bias,with a root mean square error of 1.84 m/s.This study also proves that the CMOD5.N and CMOD7 built on low and medium resolution scatterometry still have strong robustness when applied to SAR data.At the same time,the results also show that it is more important to add other geophysical parameters into the model to improve the wind speed retrieval performance in coastal areas than to increase the data resolution.In addition,with the new generation of SAR satellite-borne sensors,SAR-based surface wind speed applications no longer need to rely on the scatterometer geophysical model function.This method also opens up a new perspective for the development of HH-and compact-polarized geophysical model functions.(2)In order to avoid the backscatter conversion errors when retrieving sea surface wind speed from HH-polarization SAR images,more than 2700 ENVISAT/ASAR images are matched with ASCAT scatterometer wind products to develop the geophysical model function(CMODH)for ocean surface wind speed retrieval from C-band HH-polarization SAR images.1312 RS-2 and 107 Sl-1 satellite images are collocated with buoy observation,which is used for case study and statistical comparison in different wind speed,wind direction and incidence angle bins.The results show that the retrieved wind speeds using CMODH show a good consistency with the buoy measurements,with a statistical bias of 0.07 m/s and a root mean square error of 1.66 m/s for RS-2,and the statistical bias and root mean square error for Sl-1 are 0.49 m/s and 2.05 m/s,respectively.Compared to hybrid models,such as CMOD5.N and various PR models,CMODH modifies the wind direction bias to some extent and achieves the smallest root mean square error for wind speed retrieval.In addition,CMODH and CMOD5.N are used to evaluate the wind speed mapping capability of HH-and VV-polarization,which shows that CMODH is better than CMOD5.N for incidence angles between 30° and 49° and wind speeds between 10 m/s and 20 m/s.(3)1734 RS-2 SAR images acquired with fine quad-polarization mode are used to simulate compact-polarization data and then collocate with in situ buoy observation.This data set is used to develop and verify the geophysical model function for Canadian Radarsat Constellation Mission ocean surface wind speed retrieval.The relationship between the compact-polarization backscattering coefficient and the wind speed,wind direction and radar incidence angle is analyzed,showing that RR-polarized backscattering mainly depends on the wind speed and is less sensitive to the incidence angle and wind direction when compared with RH-,RV-,and RL-polarization.Statistical comparison shows that the retrieved wind speed using CMODRH,CMODRV,CMODRL and CMODRR is in good agreement with the buoy measurements,and the root mean square error is 1.38,1.51,1.47 and 1.25 m/s,respectively.Therefore,compactpolarization is a good alternative to linear polarization in ocean surface wind speed derivation,and RR-polarization is more suitable for high wind speed retrieval than RH-,RV-and RLpolarization modes. |
PDF Full Text Request