Benefit assessment of glider adaptive sampling in the Ligurian Sea

The benefits of piloting a glider during a 6-day period via an adaptive sampling procedure in a 80×60 km2 marine area are assessed under a fully operational framework. The glider trajectory was adapted to reduce the ocean temperature uncertainties predicted by the operational 3-D super-ensemble model in the Ligurian Sea in August 2010. Two sets of real time model predictions are compared, which assimilate observations from (1) the adaptive-sampling-driven glider and (2) an independent glider flying in the same area. The piloting algorithm was able to successfully guide the glider along the planned trajectories. These were nevertheless not fully completed due to un-predicted adverse currents faced along the transects. Despite operational constraints and model prediction errors, the adaptive sampling procedure is shown to meet the proposed objective, i.e. a reduction of the 48-h model temperature uncertainty predicted in the upper 200 m. Moreover, measurements collected during the last 48-h forecast cycle from (i) an ocean mooring, (ii) a repeated ScanFish transect and (iii) more irregularly distributed platforms, all indicate that the actual prediction error is lower in the simulation assimilating the data from the adaptive sampling. Quantitatively, the total root-mean-square error is reduced by 18% at the end of the field experiment in comparison with the control simulation.

► Adaptive sampling method merging glider mission planning and model uncertainty.
► Successful operational use of the method applied to a single glider.
► Comparison of the adaptive sampling with a different glider sampling strategy.
► Assessment of adaptive sampling performance: model forecast error reduced by 18%.