Non-synoptic versus pseudo-synoptic data sets: an assimilation experiment

Several first-order correction methods are implemented to compute pseudo-synoptic data sets from non-synoptic raw data sets. These include a geostrophic relocation method, a linear and a quadratic interpolation method, and a method using spatio-temporal correlation functions. The relocation method involves analyses and geostrophic velocity computations to allow the relocation of stations in time and space to a particular analysis time. Interpolation methods involve several almost identical and consecutive surveys interpolated in time. Temporal weighting methods are based upon a spatio-temporal function modifying the weight on data with respect to the time at which they have been sampled. These techniques are tested on the OMEGA data set and are validated by simple nudging assimilation into a 3D primitive equation model (PE). It is shown that, under certain hypothesis, these methods are able to correct the lack of synopticity in hydrographic data sets, and improve the diagnosis of vertical velocities computed from the Omega equation. These methods are of particular interest for the scientific community. They might be used together with diagnostic models. They might provide suitable pseudo-synoptic fields required by 3D PE models as initial conditions, boundary conditions or assimilation data sets. They may also be useful in the design of mesoscale samplings. (C) 2001 Elsevier Science B.V. All rights reserved.