Environmental Control of Wind Response to Sea Surface Temperature Patterns From Remote Sensing Data.

Air-sea interactions are ubiquitous but extremely diverse depending on the scale and region. At large scales, the atmosphere drives ocean responses through several mechanisms, while meso- and sub-mesoscale sea surface temperature (SST) patterns have been shown to influence surface wind convergence, atmospheric boundary layer (ABL) mixing and low-level cloud dynamics. Considerable differences emerge from region to region depending on the environmental conditions and their influence on air-sea interaction mechanisms. The dependence of mesoscale SST-wind coupling mechanisms on large-scale atmospheric features - specifically, surface wind patterns and air-sea temperature differences - has been proven using reanalysis data, but it still awaits further confirmations from observational studies at higher resolution. Several case studies show interesting interactions occurring at the sub-mesoscales, dampening or reinforcing the overlying mesoscale phenomena depending on the situation. However, a systematic assessment of sub-mesoscale air-sea interaction is still missing due to the scarcity of high-resolution observations on a global scale. A better understanding of the role played by environmental conditions at different scales would not only improve our description of surface winds but could also provide crucial information on shallow cumuli dynamics, which is a major source of uncertainty in current climate models. We aim to fill these gaps by employing high-resolution satellite observations (on grid spacing ranging from 1 km to 12.5 km) of surface wind (from ASCAT and Sentinel-1 L2 OCN product) and SST (from AVHRR) together with new products estimating atmospheric instability and surface fluxes from Sentinel-1 Synthetic Aperture Radar (SAR) data. The goal is to study the variability of SST-wind coupling intensity as a function of the environmental conditions and to determine whether the same behaviour occurs at the sub-mesoscale. We analyze 2 years of data (2020-2021) over the Gulf Stream region, as it is a region with strong SST gradients, a wide range of atmospheric stability conditions and large-scale winds up to 25 m/s. The two main SST-wind coupling mechanisms acting at the mesoscales (downward momentum mixing, DMM, and pressure adjustment, PA) are taken into consideration according to the metrics defined by Meroni (2022, 2023). In this formulation, the contributions of the two mechanisms to wind divergence are disentangled by separating the wind and SST gradient components along (DMM) and across (PA) the background wind direction. The intensity of the coupling between these gradients is evaluated as a function of the large-scale surface wind speed and a few proxies of atmospheric instability, that have also been shown to modulate shallow convective phenomena. Estimates of atmospheric instability (namely Obukhov length and buoyancy flux) are retrieved from Sentinel-1 SAR images through a machine-learning algorithm (O’Driscoll et al. 2023). The scenes are acquired in WV mode, the default acquisition mode over the ocean and consist of 20 km by 20 km vignettes sampled roughly every 100 km along the orbit. Only scenes containing convective conditions are considered. These SAR-derived environmental conditions are compared with air-sea temperature difference and buoyancy flux from hourly ERA5 data to determine whether or not their use contributes to a more appropriate characterization of the environmental conditions and as a consequence of the air-sea interactions themselves. The large-scale background wind, strictly related to the ABL mixing and friction velocity, is also included as an environmental condition, and it is obtained by low-pass filtering MetOp A ASCAT wind data. A control analysis conducted on the MetOp database before the colocation with SAR data has shown that the intensity of DMM coupling peaks for background winds around 8 m/s and near-neutral atmospheric conditions (estimated from ERA5 air-sea temperature difference and buoyancy flux). These results confirm the well-known nature of the DMM mechanism, which is enhanced by the mixing induced by the entrainment of momentum from the troposphere into the ABL. The use of Obukhov length estimates from ERA5 as a proxy for atmospheric instability yields no significant signals in the coupling intensity. In fact, Obukhov length describes the balance between the surface stress turbulent production and the buoyancy flux, equating situations of strong instability and strong winds with situations of marginal instability and weak winds which nonetheless exhibit significantly different mixing dynamics. The same methodology has been applied to the database of co-located MetOp A and Sentinel-1 observations, with two major drawbacks. First, a drastic reduction in the amount of data, which reflects in a scarcity of strong SST gradients and therefore in an overall weakening of the coupling signal. Second, as the SAR database contains only shallow convection scenes, the sensitivity to atmospheric stability is very limited. In this case, DMM is more active in situations of high wind speed (around 12m/s) and near-neutral atmospheric conditions, but with some concerns regarding the robustness of the detected signal. Both the control and co-located analyses conducted on the PA mechanism have shown no significant results when aggregating the data according to the above-mentioned environmental conditions. The reason behind this behaviour is likely to be found in a generalized scarcity of strong SST laplacian needed for PA to develop, suggesting the need for a larger database. To effectively employ SAR-derived estimates in the study of SST-wind interactions it is necessary to expand the analysis beyond the limits imposed by the current SAR database. To this end, a new dataset of SAR observations from Sentinel-1 SAR Interferometric Wide (IW) Swath mode scenes is currently under development. The IW mode is not the default acquisition mode over the ocean, but it is often sampled near coastal areas (where most of the WBCs and the associated SST patterns lie) and major islands (as in the North-West-Atlantic Tropical region). IW scenes consist of continuous swaths of 200 km width, providing a much larger amount of data compared with the sparse 20 km by 20 km images of the previous WV mode. Finally, the availability of Sentinel-1 L2 OCN product, containing surface wind at 1 km resolution, allows to extend the analysis to the sub-mesoscales, once coupled with an appropriate SST product. Since SAR wind retrieval is strongly dependent on wind speed and direction, this step requires an extended investigation of the bias and limits introduced by the use of such a product. The challenges posed by this new methodology are of major interest in the path toward the deployment of the Harmony constellation, as they provide useful insight into the potential of the new observations that the mission will make available to the scientific community.