Time series analysis and prediction of multi-dimensional signals in geodesy
Convener: Wieslaw Kosek  | Co-Convener: Michael Schmidt 
Oral Programme
 / Tue, 04 May, 08:30–10:00  / Room 7
Poster Programme
 / Attendance Tue, 04 May, 17:30–19:00  / Halls X/Y

Observations of space geodetic techniques (geometric and gravimetric) deliver a global picture of the dynamics of the Earth usually represented in the form of time series which describe (1) changes of the surface geometry, variations of the ocean surface and ice covers, (2) the fluctuations in the orientation of the Earth and (3) the variations of the Earth’s gravitational field. However, the temporal variations represent mostly the total, integral effect of all mass exchanges between all elements of the Earth’s system including atmosphere, ocean and hydrology. Different time series analysis methods have to be applied to analyze all these geodetic time series for a better understanding of the relations between the elements of the Earth’s system as well as their geophysical causes. The interactions between the different components of the Earth’s system are very complex, irregular and non-stationary. Thus, it is necessary to apply time frequency analysis methods in order to analyze these time series in different frequency bands as well as to explain their relations to geophysical processes. Consequently contributions to time frequency analysis and representation to display reliably the features of the temporal or spatial variability of signals existing in various geodetic data, as well as in other data sources such as geophysical models are highly appreciated.
Time series analysis methods can be also applied to analyze multi-dimensional data on the surface including maps of the gravity field, sea level or atmospheric signals as well as temporal variations of such surface data. Many one-dimensional techniques can be generalized easily to the multi-dimensional case. We appreciate contributions on efficient generalizations of appropriate analysis procedures. Another important problem concerns the estimation of deterministic (including trend and periodic variations) and stochastic (non-periodic variations and random changes) components of different geodetic and geophysical time series as well as the application of digital filters for extracting specific components with a chosen frequency bandwidth. The multiple methods of time series analysis may be encouraged to be applied to the preprocessing of raw data from various geodetic measurements in order to promote the quality level of enhancement of signals existing in these raw data.
We further expect contributions related to specific issues such as the improvement of edge effects, since they may affect the reliability of long-range tendency (trends) estimated from data series. We solicit papers on different prediction techniques e.g. least-squares, neural networks, Kalman filter or uni- or multivariate autoregressive methods to forecast Earth Orientation Parameters, which are needed for real-time transformation between celestial and terrestrial reference frames.

Related event: G6 – Space geodetic techniques and the Earth's atmosphere
Oral Programme
 / Thu, 06 May, 13:30–17:00  / Room 19
Poster Programme
 / Attendance Thu, 06 May, 17:30–19:00  / Halls X/Y