Improving tide-gauge data processing: A method involving tidal frequencies and inverted barometer effect

Sea level measured at tide-gauge stations is affected by undesired events such as spikes, gaps, steps and time shifts. The presence of these instrumental failures seriously decreases the quality of each analysis done using the record. Furthermore, they can mask or disturb oceanographic phenomena of small amplitude. Identifying and filling gaps in the records have as consequence longer data sets, allowing the study and discrimination of long-term components. In this work, a method included in a reduction software package is presented. It is based on the oceanographic characteristics of the time series, mainly ocean tide and inverted barometer response. Raw data are interpolated using polynomial least-squares approximation, considering variable parameters, i.e. degree and number of used data, estimating the optimal ones by contrast tests. At the same time, estimation of errors and maximal length of the gaps to be interpolated are obtained. The determination of these parameters is done using Least-Squares Techniques. Since the vertical displacement of the oceanic surface due to atmospheric pressure variations is of considerable magnitude, the study is extended to the elimination of this effect prior to the interpolation. The preceding results, i.e. the adjusted parameters, are used to build up an error detection procedure based on classical polynomial interpolation, which has been programmed in Fortran 90. The correction is done at the residuals' level, meaning that estimated minus observed values. The user has full control over the iterative adjustment process. Finally, the obtained results are compared with the results provided by other pre-processing methods.