Estimativa de vento utilizando imagens SAR do satélite Sentinel-1 da ESA

Abstract

Through remote sensing it is possible to obtain information about Earth's surface without direct contact with it. The process of data acquisition involves interactions between the incident radiation and the target of interest. The energy required for remote sensing is the electromagnetic radiation (EMR), which has as its most important characteristics its frequency (υ) and its wavelength (λ). Active sensors send out energy and then measure the properties of the returning signal after it was absorbed, reflected or spread over the target's surface. Synthetic aperture radars (SAR) allow increased spatial resolution of the image obtained from the ground regardless of platform altitude and generate two-dimensional images of high resolution independent of daylight, cloud cover and weather conditions. These systems are the only way of measuring wind on the water surface with spatial resolution of a few hundred meters. Thus, this study aims to estimate the wind fields on the ocean surface using Sentinel-1 images, a polar orbiting satellite launched in April 2014 by the European Space Agency (ESA), which has a synthetic aperture radar C-band (CSAR). To achieve this goal, images of the Bazilian coast obtained by Sentinel-1 and acquired from ESA, were treated and analyzed using the SNAP 2.0 software. Then, the wind speed and direction on the surface of the sea were estimated from these images and compared with GFS model data. The correlation coefficients obtained between the velocities of the model and of the images for January 27th , February 6th and February 20th were -0.047, -0.541 and 0.684 respectively. Regarding wind direction, there was a good similarity between the two types of data only for January 27th . In conclusion, the techniques of estimation of the wind field (velocity and direction) used for images of Sentinel- 1 showed efficient results and their acquired data have presented a significant correlation with the data provided by the GFS model to two out of the three images that were used.