Abstract
This internship project performed a quality analysis of Sentinel-1 SAR-derived wind speeds under critical atmospheric conditions encountered off the U.S East Coast, focusing on a routinely covered Sentinel-1 SAR frame in which the northwestern boundary of the Gulf Stream isvisible in many images acquired within a one-year time period. This frame was selected in order to find out if the difference between SAR-derived and reference wind speeds changes with the air-sea temperature difference. Winds over the Gulf Stream are often associated with an unstable atmospheric condition because the water is warmer than the air on average, while the atmospheric stability outside the Gulf Stream can be expected to be more neutral. Wind retrieval algorithms for wind scatterometer data and SAR images are usually tuned to produce accurate results under neutral stability conditions. An analysis on the dependence of theSAR-derived wind speed bias on the air-sea temperature difference (T) was performed, including a statistical analysis to quantify the extent of over-or under estimation of Sentinel-1 SAR-derived wind speeds relative to ERA5 reanalysis winds from the European Centre for Medium Range Weather Forecasts (ECMWF). Interestingly, we found a relatively large underestimation of wind speeds by the SAR wind algorithm under neutrally stable conditions and a smaller overestimation under unstable conditions. This confirms the expected quantitative trend, but with an unexpected bias towards better agreement under unstable conditions. The wind speed bias was found to depend on the wind speed itself as well, but the trend in the dependence on Tis significant. It is visible in our statistical results for all data combined, but also in monthly and seasonally averaged wind bias and T maps that show clear differences between the areas over and outside the Gulf Stream. However, the differences across the Gulf Stream front are more pronounced during winter and spring than during summer.The small wind speed overestimation over the Gulf Stream by the SAR wind algorithm suggests a good quality of the Sentinel-1 SAR-derived wind speeds under unstable atmospheric conditions. The under estimation under neutral atmospheric conditions appears to be a systematic bias of the wind retrieval algorithm, but this finding may be specific to the test area considered here, which is relatively close to the coast. Extended studies with more data from other regions would be required to study the nature of this general bias in full detail.