Some aspects of geodetic time series modelling
The radical technological development of sensors, communication techniques and high-performance computers has increased the importance of permanent fully automatised monitoring systems, also geodetical, and promoted, encouraged or even required the development and testing of enhanced modelling algorithms in order to extract all of the information from the data available. The implementation of mathematical techniques prioritizes the models which enable the assimilation of all of the information available, an estimation of essential parameters of dynamic modelling in real time, the processing of measurements with data gaps and methods to handle the complexity of adopted functional models. As such these models can provide an additional supportive tool for safety assessment with decision-making and can distribute more reliable results for alarming procedures. In the contribution an overview of some methods, which were presented in detail in referenced manuscripts, is given. Long-term time series of monitoring data were used to implement and analyse regression analysis and Bayesian probabilistic approach for model class selection, two models of Kalman filtering for (near) real-time assessment, and two frequency methods - Least Squares Spectral Analysis and Lomb-Scargle normalised periodogram, for analysing periodicities in time series with unevenly distributed data.