Space Research Centre, Bartycka 18A, 00-716 Warsaw, Poland
(*) On leave from Institute of Applied Astronomy, St.Petersburg, Russia
Global geophysical processes which excite variations in Earth rotation contain a significant irregular constituent of random nature, which indicate that a stochastic approach to description of polar motion is the most appropriate. The stochastic model of the Chandler wobble, which is considered in this paper, goes back to Jeffreys (1940). This model is derived from the equation of polar motion, besides our statistical tests on the observed polar motion series prove its adequacy and superiority over certain deterministic models applied so far. In practice, the Jeffreys model can be incorporated through its covariance function into such stochastic methods of data processing as the Kalman filter and the least squares collocation technique. An important problem of geophysical interest, which can be solved by the use of these procedures is deconvolution of the polar motion series yielding the excitation function. Other tasks, e.g. smoothing and prediction of the polar motion data, are also within the scope of these algorithms though they are mostly technical problems. Possible extensions of the Jeffreys model are considered to cover all other components of polar motion which show a random behaviour, such as seasonal variations and rapid oscillations with periods of the order of weeks.