Figure 3: Synthetic mean velocities in the Mid Atlantic Bight region off New Jersey and Delaware, estimated (a) from drifting buoy (at 1/8°) and (b) from HF radar data. Black lines represent 100m and 2000m isobaths.

2.4 Multivariate objective analysis

The third step of this method is multivariate objective analysis as in Rio and Hernandez (2004), Rio et al. (2007, 2011 and 2014a) and in Mulet et al. (2021), which uses the synthetic mean geostrophic velocities and synthetic mean heights to improve the first guess, in particular to improve the fine scales, to obtain the CNES-CLS22 MDT. This optimal analysis requires the a priori MDT variance and the a priori zonal and meridional spatial correlation scales of the estimated field. The same statistical a priori as for Rio et al. (2014a) and Mulet et al. (2021) are used here. In the equatorial band, as the geostrophic approximation is no longer valid, only mean synthetic height observations are used for MDT estimation, and for current estimation, only mean synthetic velocities observations are used for inversion.