An assessment of marine atmospheric boundary layer roll detection using Sentinel-1 SAR data

Abstract

The ability of high-resolution synthetic aperture radar (SAR) to detect marine atmospheric boundary layer (MABL) roll-induced roughness modulation of the sea surface wave field is well known. This study presents SAR measurements of MABL rolls using global coverage data collected by the European Space Agency's C-band Sentinel-1A satellite in 2016-2017. An automated classifier is used to identify likely roll events from more than 1.3 million images that were acquired at two incidence angles of 23° and 36.5° in either VV or HH polarization. Characteristics of the detected rolls are examined for different wind speeds, polarizations, incidence and relative azimuth angles. Roll detection counts are much higher at the higher incidence angle and nearly equivalent for VV and HH polarizations. Detection depends strongly on the relative azimuth with roll detection rates at crosswind being 3-10 times lower than for up-or downwind. All data show a low wind speed threshold near 2 m s−1 and that rolls are most commonly observed at wind speeds near 9 m s−1. For all viewing configurations, we find that rolls induce a wide range of mean surface wind speed modulation with the most frequent value being 8% (±3.5%). Roll detection at crosswind is associated with stronger roll-induced surface wind enhancement. Dependencies of roll detection on the incidence and relative azimuth angles are consistent with rapid short-scale wind-wave adjustments to the roll-induced surface wind gusts. These cm-scale waves are highly directional and provide limited crosswind backscatter at shallower incidence angles. The same roll-induced surface forcing is thus not equally detectable at all viewing geometries or polarizaions. Stronger and possibly longer-duration wind forcing is likely needed to produce detectable roll-induced modulations at crosswind. Highlights► First global statistics of SAR response to MABL rolls using new S-1 WV data. ► Data show 50% greater sensitivity to rolls for 36° vs. 23° SAR incidence angle. ► Crosswind (relative to SAR look direction) shows weakest sensitivity to roll imprints. ► Rolls observed at 23° and crosswind are more stationary, in more unstable conditions. ► SAR-estimated surface wind perturbations due to roll impacts are 8 ± 3.5%.

Multi-Scale LG-Mod Analysis for a More Reliable SAR Sea Surface Wind Directions Retrieval

“…Wind rows are detected mostly in the case of light-moderate wind speed regimes (i.e., W less than 13.8 m/s) and along-shore winds (Figure 17a,b, respectively). This finding confirms the results in References [20,21], where it is reported that wind rows are most commonly observed at wind speeds near 8-9 m/s. Whatever the wind speed and direction, the Low-and the Medium-Scale selections occur more frequently than the High-Scale one.…”

“…The developed algorithms basically rely upon the estimation of the prevailing local orientation of linear wind induced and aligned SAR signatures, such as those from Boundary Layer Rolls (BLRs) and Wind Streaks (WSs) [18,19]. These features, otherwise termed "wind rows", could be visible on SAR amplitude in favorable conditions, which are determined by thermal and dynamic air-sea instability and typical wind speed values from 2 m/s, with the highest probability of wind streaks in the SAR images found for wind speeds of about 8-9 m/s [20,21].…”

Automated Rain Detection by Dual-Polarization Sentinel-1 Data

“…The dual-polarization filter is hopefully applied on other C-band satellite data, i.e., RADARSAT-2, yet the incidence angle differing from Sentinel-1 might gave slightly different performance. The encouraging performance of a dual-polarization filter to detect rain signatures in SAR images seems to be applicable also to other, non-rain-related features visible on SAR images, such as roll vortices [2]. Additionally, in the particular case of rain detection and segmentation, further work could be done to decipher between the various rain contributions (i.e., stratiform rain, convective rain) and the possible relationship of backscattered signal with lighting events by exploiting the full capability of SAR (for instance, by including phase information), including data from ground-based radars but also other ancillary products such as the Geostationary Global Lightning Mapper (GLM).…”

“…Copernicus level-2 ocean products encompasses waves (OSW), wind (OWI) and radial surface velocities (RVL) measurements. However, many other ocean and atmospheric geophysical phenomena such as oceanic front, convective cells, modulation of the stress in the marine atmospheric boundary layer or rain [1,2] also affect the signal. Although routinely available since the Radarsat-2 mission, the use of dual-polarization measurements for ocean processes analysis and level-2 ocean product definition is still emerging.…”

Semantic Segmentation of Metoceanic Processes Using SAR Observations and Deep Learning

“…It was later supplemented by another 10,000 images. In [14], deep learning models trained from this dataset led to promising results with an excellent classification performance [15]. However, this effort could not generalize to imagettes with at least two phenomena, and, as a categorization task, is inherently limited in resolution.…”