Level 2 Ocean Wind Field Processing
The SENTINEL-1 Level-2 OWI component is an ocean surface wind vector (speed and direction) estimated from a SENTINEL-1 Level-1 SAR image by inversion of its associated Normalised Radar Cross Section (NRCS).
SENTINEL-1 wind processing consists of a calibration unit followed by an inversion unit.
The Level-2 OWI processing system will access:
Calibration and Pre-processing
The calibration unit performs the processing from Level-1 product to a calibrated product containing the parameters required for the wind inversion unit. They are calculated at the resolution of the desired wind cell for each wind cell of the OWI grid.
Given the size in metres, of the SAR derived wind cells (indicated in the Level-2 processor parameters file) and the pixel size in metres (such as read in the data Level-1 header), the parameters for the bright target detection, the number of pixels to be averaged in range (x) and azimuth (y) direction are computed. Based on this, the parameters of the OWI are estimated. Arrays corresponding to the intensity image, NRCS, longitude and latitude, incidence angle, track angle, two-way slant range time and the percentage of pixels in SAR image detected as bright targets are generated.
The a priori wind and ice mask information from external ancillary data are interpolated on the SAR OWI grid. The coastline and land mask information from internal ancillary data are interpolated on the SAR OWI grid. The calibration constant available in the Level-1 product is used as external input data. In addition, an internal calibration constant based on ancillary geophysical parameters is calculated for each SAR image product.
The inversion unit generates the Level-2 OWI component using parameters obtained from the calibration unit. The inversion estimates the wind vector using NRCS values, the incidence angles, the track angles and the a priori wind and ice information. This inversion is performed using statistical Bayesian inference. For HH polarisation, the NRCS in HH is transformed to a VV equivalent NRCS. For dual polarisation data, the cross-polarisation NRCS is not used in the wind inversion scheme. The possible wind speed values at each wind cell are estimated given the NRCS value and all possible wind directions. At each wind cell, the probability density function of a priori wind vector is computed for all the possible solution couples. Next, the most plausible wind speed and direction values at each SAR wind cell of the SAR wind grid are selected. Lastly, the SAR and ancillary model information consistency is estimated. The OWI component is then saved to be joined with the other components into a netCDF file.