Sea-level signature of bathymetric errors and their observability by satellite altimetry
Baptiste Mourre(1) , Pierre De Mey(2) , Florent Lyard(2) , and Christian Le Provost(2)
Passeig Maritim, 37-49,
(2) LEGOS, 18, av Edouard Belin, F-31401 Toulouse Cedex 9, France
An experiment of bathymetry correction via the assimilation of sea level altimeter observations in a 2-dimensional numerical hydrodynamic model implemented on the North Sea shelf is presented here. Estimating sea level model error statistics due to uncertainties in bathymetry, and their relation to bathymetric errors, effectively makes it theoretically possible to update the bathymetric solution using both sea level observations and data assimilation techniques. In this work, this estimation is empirically performed making use of ensemble methods.
A single experiment is carried out, as a by-product of a more general study aiming at estimating the ability of different altimeter configurations to control model error on the North Sea shelf. This is the reason why we do not aim at bringing here a complete answer to the question but rather propose a single illustration about the topic. We use the barotropic MOG2D model, and the focus is on the specific high-frequency response of the ocean to meteorological forcing (time scales from a few hours to a few days).
Sea level model error due to uncertainties in bathymetry is first carefully explored. An ensemble of bathymetric solutions is generated by randomly combining over the study domain elementary perturbations extracted from typical mismatches between different existing bathymetric databases. The model run over these perturbed solutions leads to an ensemble of simulations enabling to approximate model error statistics related to this particular source of error, and therefore to give an insight into the signature of bathymetric errors on model sea level. This signature appears to be neither homogeneous over the shelf, nor stationary, since it strongly depends on the meteorological regime and the associated oceanic processes at work.
Seafloor elevations are then included in the state vector together with model sea level, in order for them to be updated during the analysis. A space-time global inversion scheme taking into account all observed variables over a 20-day period is implemented. In the framework of twin experiments, we consider simulated high-resolution WSOA data (Wide Swath Ocean Altimeter) to correct a given perturbed bathymetric solution along the eastern British coast.
Results are encouraging in the sense that the bathymetric update approximates to the control solution. However, a large part of the error remains uncorrected, meaning that bathymetric error signature on model sea level is not significant enough in this particular experiment to be observed by the particular altimeter observations under consideration. A more thorough study would be needed to confirm these first findings.