Advanced Time Series Models
Benudhar Sahu · 2021
This chapter discusses a few more advanced and powerful modelling techniques for time series data analysis, forecasting and control. A structural time series model, such as classical decomposition, is specified in terms of components: trends, seasonality and noise, which are of interest themselves. Gaussian linear processes have characteristic time-reversibility or symmetry, but the observed time series may be asymmetric. There has been growing concern about the discrimination of natural earthquakes versus artificial explosions since in the former case, certain hazard precautions and mitigation efforts have to be planned. The seismic recording stations have a continuous record of P and S wave arrivals along two horizontal and vertical directions so that amplitudes and phases are discernible. The amplitude ratios of P and S waves may be less for natural earthquakes as compared to artificial explosions, since the former is for long-time periods, whereas the latter is for very short-time period.