Sentinel 1 (Continuation)

The Sentinel -1 Level-2 WV OCN product contains three geophysical components: the radial surface velocity (RVL), the ocean surface wind field (OWI) and the ocean swell wave spectra (OSW). The main difference between the former ASAR WM Level 2 product and S1 OSW component is higher range bandwidth (74.5 MHz versus 16 MHz) and larger imagettes (20x20km²) versus 5x5km²) allowing improved detection of very short waves and very long swell systems. In addition, the S1 OCN product contains a RVL component providing estimates of Doppler centroid (DC) frequency and the corresponding radial velocity. The S1 Level 2 WV OCN product contains also co-located wind speed and direction from the global atmospheric model of ECMWF. As part of the Sentinel-1 Mission Performance Center activities, an automatic collocation system is set-up at Ifremer providing collocated wave model (WW3), scatterometer (ASCAT-A, ASCAT-B) as well as altimeter (Cryosat-2) data in near real-time and on global basis. Collocation with buoys is also undertaken. A flexible reprocessing system including the S1 IPF Level 2 processor is installed at the cluster at Ifremer. The system is capable of reprocessing of one month of S1 WV OCN products within few hours. The system is used to upscale and evaluate on massive data, evolution of the Level 2 processing algorithms.

The operational 2D ocean wave spectra of the OSW product is validated partition by partition against WW3, and also for effective H_s. For effective H_s a standard deviation of around 0.45 m is achieved for WV1 (incidence angle 23^o), while around 0.55m for WV2 (incidence angle 36^o). In general the WV1 performs better than WV2 due to larger SNR. The validation results show that the main deviations are related to none-wave low frequency signal contaminating the SAR wave spectra. These are likely to appear at low winds.  As of this improved filtering of low frequency signal is implemented and impact is assessed using the reprocessing capacity at Ifremer. The Level-2 algorithm improvements and upgrades will be part future S1 IPF releases.

The OCN products from Sentinel-1 routinely produced by the PDGS and circulated to the users consist on 3 main components: surface wind retrieval (OWI), radial velocity (RVL), swell (OSW) [1]. This conference paper is focused on the validation and the status of performance for the OWI component, to be shared with the Sea Sar community.

First, it will be reminded the strategy put in place to validate those products at Mission performance center. In a second part, we will summarize the performance status of those products (past and present for the Sentinel 1 missuion). Then, it will present the short-term evolution for the OWI products.

Strategy of validation

The validation of these products is made as a collaboration between CLS and Ifremer, in the framework of Mission Performance Center. The strategy to assess the accuracy of the wind retrieval is to compare it with an auxiliary wind source which is used as a reference. This source could be in-situ data from buoy, other satellite data (ex: scatterometer) or atmospheric model outputs. It is important to outline the importance to multiple the type of data used as reference, due to their coverage, resolution or possible bias. In this scope, Ifremer has performed systematic collocations with such data with L2 products generated by the ESA-IPF by PDGS.

Status of performance

It can be noticed the strong correlation of the SAR-derived wind speeds with the wind references. The bias and the RMS are less important for the comparison with ECMWF re-analysis wind model, this is expected since the wind inversion is based on the ECMWF forecast as an a priori wind input. A nominal RMS of 1.5m/s to 2m/s is observed, for all the wind reference. A bias of around 0.3 to 0.7 m/s depending on the wind reference is present. This bias may be related to the choice of the current geophysical model function (GMF) used for the wind retrieval. As expected, at low wind speeds, the NESZ impacts the SAR wind measurement. Then, a noticeable impact of the noise can be observed on HH polarization and high incidence angle (over estimation, beam shaped profile).  The performances are similar for both satellites. Some work to improve the performance are currently conducted and will be presented.

IPF2.90 Content, short term evolution

With the next version of the processor IPF 2.90, the format of L2 data has been extended in order to include additional information on cross-polarization (NRCS, NRCS corrected and NESZ annotation on the OWI grid) [4]. This is a good opportunity for the community working on very strong wind-speed conditions (extreme wind phenomena: ex hurricanes…). The products that will be processed with this IPF version will also contain the new NESZ annotation (variations on the azimuth).

We would like by this paper take the opportunity to share with to the user Community the strategy of the validation of the OWI component of Sentinel-1 OCN products, the status on their performances (present and past), and the short-term evolution planned for the next year.