Time series analysis: forecasting and control.

Abstract

We perform a statistical analysis of measured time series of sea levels at various coastal locations in the UK, measured at time differences of 15 min over the past 20 years. When the astronomical tide and other deterministic components are removed from the record, a stochastic signal corresponding to the meteorological component remains, and this is well-described by a superstatistical model. We do various tests on the measured time series, and compare the data at 5 different UK locations. Overall the χ2-superstatistics is best suitable to describe the data, in particular when one looks at the dynamics of sea-level differences on short time scales.