Validation of Sea Surface Wind From Sentinel-1A/B SAR Data in the Coastal Regions of the Korean Peninsula

Abstract

In this study, using in situ measurements at 17 buoy stations off the Korean Peninsula, C-band model (CMOD) functions for Sentinel-1A/B IW mode synthetic aperture radar (SAR) data were validated. In total, 395 Sentinel-1A/B IW mode dual-vertical polarized images were used for collocation with in situ measurements from May 1, 2015, to September 30, 2017, and 807 matchup points were obtained. Prior to the validation, preprocessing such as speckle noise reduction and ship and land masking was completed. The in situ wind speeds were converted to a 10-m neutral wind considering atmospheric stability. High-resolution wind speeds were estimated by using the CMOD functions such as CMOD4, CMOD_IFR2, CMOD5, CMOD5.N, and CMOD5.Na. The root-mean-square errors of each model were less than approximately 1.8 m·s–1 (1.83, 1.82, 1.69, 1.68, and 1.65 m·s–1, respectively). The biases of all models were higher in the western coastal region than those in the eastern coastal region. The results showed the advantages and disadvantages of each model in the estimation of wind speeds in the coastal region around the Korean Peninsula as proposed in a concept of combined errors. The wind speeds derived from the SAR data also presented a tendency for water depth to be overestimated over shallow bathymetry and to be underestimated at high wind speeds. In addition, this study assessed potential sources of wind speed errors such as the effects originating from wind direction input, different platforms of Sentinel-1A and Sentinel-1B and their calibration, and from radar interference or regional oceanic characteristic environments.