A Novel Method to Retrieve Significant Wave Height from GF-3 Quad-polarization SAR Imagery

Résumé fourni par la source

It has been long accepted that ocean wave states parameters can be well acquired from SAR aboard satellite. Comparing with NRCS, the effect of cutoff caused by radial velocity motion showed highly correlation to SWH and potential for SWH retrieval from SAR imagery. Following the scheme of${{\rm{\lambda }}}_{\mathrm{c}}$estimation from Chinese GF-3 SAR imagery, the resolution of sub-looks and polarization selection were compared for cutoff wavelength calculation. It is revealed that, for HV polarization, estimated${{\rm{\lambda }}}_{\mathrm{c}}$are generally larger than HH ones, the latter already being larger than values estimated in VV polarization. Especially, to propose an objective way to filter unreliable${{\rm{\lambda }}}_{\mathrm{c}}$, the parameters of kurtosis and skewness are related with unreliable imagery and estimated${{\rm{\lambda }}}_{\mathrm{c}}$. Moreover, the new quality criterion was performed to establish an empirical method for SWH retrieval under non-fully/fully developed sea states for HH, HV, VV polarization. The systematic comparisons showed the potential of${{\rm{\lambda }}}_{\mathrm{c}}$estimation for sea states condition, especially for non-fully developed sea states. In particular, a novel method to retrieve SWH was proposed in consideration of imagery features$( {{\sigma }_0,cvar} )$and estimated${{\rm{\lambda }}}_{\mathrm{c}}$by following the steps focused in this study. The independent validation data provided additional sea state information related to estimated${{\rm{\lambda }}}_{\mathrm{c}}$. Validation against SWH from ECMWF showed that the performance of$f( {{\sigma }_0,cvar,{\lambda }_c} )$showed satisfying results, with a Cor of 0.77 (HH), 0.73 (HV), 0.77 (VV), and a RMSE of 0.18 (HH), 0.19 (HV), 0.18 (VV). This demonstrates that this method$f( {{\sigma }_0,cvar,{\lambda }_c} )$can well develop retrieval precision of SWH.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.