Relative performances of WSOA, altimeter constellations and tide gauges in controlling a model of North Sea barotropic dynamics
Baptiste Mourre(1) , Pierre De Mey(2) , Florent Lyard(2) , Yves Ménard(2) , and Christian Le Provost(2)
Passeig Maritim, 37-49,
(2) LEGOS, 18, av Edouard Belin, F-31401 Toulouse Cedex 9, France
The relative abilities of different possible future altimeter systems to control model error due to uncertainties in bathymetry in a barotropic model implemented on the North Sea shelf are estimated in this work. The focus is on the specific high-frequency response of the ocean to meteorological forcing, involving temporal scales from a few hours to a few days. Due to the specificities of the shelf dynamics, a special attention is paid to properly specify the shape and evolution of model error statistics, that are needed to assimilate sea level data into the model. Ensemble methods are used for this purpose, and model error due to uncertainties in bathymetry is more specifically studied. The finite-element barotropic model MOG2D is run over an ensemble of perturbed bathymetric solutions to approximate error covariances. An ensemble Kalman filter is then implemented to assimilate data in the model.
We performed multiple twin experiments to assess the capability of different observing scenarios to reduce model error (in the framework of so-called Observing-Systems Simulation Experiments). The diagnostic is based on the reduction of the ensemble spread thanks to the assimilation. The contributions of the Wide Swath Ocean Altimeter and different satellite constellations are first investigated.
In the context of North Sea dynamics, a single WSOA has similar performances than 2 nadir satellites in terms of sea level correction, and is better than 3 satellites in terms of model velocity correction. Due to the short temporal scales of the particular oceanic processes under consideration, the temporal resolution of observations is shown to be of major importance for controlling model error in these experiments. Such altimeter systems are precisely shown to lack temporal sampling to properly correct the main part of model error here.
Tide gauges, that provide the required very fine temporal resolution, are then considered. In this particular context, they lead to very good global statistical performances. Looking into further detail, tide gauges and altimetry are demonstrated to exhibit some interesting complementarity, since high-resolution altimeter systems are more efficient on the inside of the shelf, whereas tide gauge networks make it possible to properly control model error in a ~100 km coastal band.