Estimating surface velocities from satellite data and numerical models: Implementation and testing of a new simple method

A simple method of fusing tracer observations and model outputs for computing surface velocities in the ocean is implemented and tested in the framework of the twin experiment approach. Synthetic data from realistic velocity outputs produced by the operational Mediterranean Forecasting System (MFS) are used. The method (Piterbarg, 2009) allows to estimate a velocity field using two consecutive tracer snapshots. The focus is on testing realistic time intervals between snapshots and partial tracer observations. The considered configuration consists of a tracer patch released and advected by the current, and is motivated by the practical problem of estimating velocities and concentrations using satellite data in case of pollutant releases such as oil spills. An extensive set of experiments has been carried out, and the method performance has been quantified in terms of improvements in accuracy with respect to the model. The improvement ranges from values of approximately 80-90% for concentration and 50-60% for velocity in the case of almost perfect data, to values of 30-40% for realistic time intervals of the order of days and reduced tracer information, and values of 15-20% when only the boundary of the patch is observed. The results are found to be robust to flow variability and patch parameters.