Altimetric sampling and mapping procedures induce spatial and temporal aliasing of the signal – characteristics of these aliasing effects in the Mediterranean Sea

Abstract. This study deals with spatial and temporal aliasing of the sea surface signal and its restitution with altimetric maps of Sea Level Anomalies (SLA) in the Mediterranean Sea. Spatial and temporal altimetry sampling, combined with a mapping process, are unable to restore high-frequency (HF) surface variability. In the Mediterranean Sea, it has been shown that signals whose intervals are less than 30–40 days are largely underestimated, and the residual HF restitution signal contains characteristic errors which make it possible to identify the spatial and temporal sampling of each satellite.

The origin of these errors is relatively complex. Three main effects are involved: the sampling of the HF long-wavelength (LW) signal, the correction of this signal's aliasing and the mapping procedure.
– The sampling depends on the characteristics of the satellites considered, but generally induces inter-track bias that needs to be corrected before the mapping procedure is applied.
– Correcting the aliasing of the HF LW signal, carried out using a barotropic model output and/or an empirical method, is not perfect. In fact, the baroclinic part of the HF LW signal is neglected and the numerical model's capabilities are limited by the spatial resolution of the model and the forcing. The empirical method cannot precisely control the corrected signal.
– The mapping process, which is optimised to improve restitution of mesoscale activity, does not propagate the LW signal far from the satellite tracks.
Even though these residual errors are very low with respect to the total signal, their signature may be visible on maps of SLAs. However, these errors can be corrected by more careful consideration of their characteristics in terms of spatial distribution induced by altimetric along-track sampling. They can also be attenuated by increasing the altimetric spatial coverage through the merging of different satellites.

Ultimately, the HF signal, which is missing in maps of SLA, can be completed using a numerical model in order to estimate the total surface signal. The barotropic HF (<30 days) component accounts for nearly 10% of the total variability. Locally, it contributes nearly 25% of the total variance.