Comparison of C-band SAR algorithms to derive surface wind vectors and initial findings in their use in marine search and rescue

Abstract

This paper presents a comparison of existing algorithms to derive surface winds from synthetic aperture radar (SAR) satellites and investigates their use in drift forecasting for search and rescue purposes. Specifically, SAR-derived winds from RADARSAT-1 and ENVISAT advanced synthetic aperture radar (ASAR) data at 1.5 km resolution are compared with scatterometer-derived winds. Three approaches were used to retrieve the wind vector from the SAR data, including an optimal inversion method combining SAR data and background numerical weather prediction, the geophysical model function CMOD-IFR2 with an a priori wind direction, and a technique that uses the backscatter values corresponding to two neighboring subimages with slightly different incidence angles. Our comparisons of SAR wind mapping with scatterometer winds from QuikSCAT and ERS-2 produced a root mean square error (RMSE) of 1.5 m/s. The optimal inversion method seems very promising and appears to be the best choice for assimilation of SAR-derived winds into operational wind products with respect to the datasets presented here. Additionally, the suitability of SAR imagery for search and rescue operations is reviewed. It is recommended that a method should be explored to automatically assimilate such data into operational search and rescue tools. Use of SAR winds in a search and rescue drift model is shown herein to produce improved drift trajectories on a number of search and rescue targets (e.g., life boat, sail boat, person in water).