Stochastic modeling of Earth's rotation change and polar motion

The rotation of the Earth exhibits relatively small variation in its rate and in the direction of its axis. The short-term changes (variations over days to years) in the rotation rate (or length of day, LOD) and polar motion (PM) are known to be strongly correlated with time-variability in distribution of mass and momentum in the atmosphere and ocean, and these relations can be approximated well as simple stochastic differential equations. Such stochastic equations have been shown useful for estimation and prediction of LOD and PM based on atmospheric and oceanic parameters (e.g., prediction of LOD using numerical forecasts of atmospheric angular momentum). The stochastic model can also be the basis of using the LOD and/or PM as signatures of some atmospheric or oceanic phenomena (e.g., possibility of verifying model-generated deep ocean circulations through ocean angular momentum). In recent years, the accuracy in the measurements of LOD and PM as well as in the numerical circulation models of atmosphere and ocean has undergone steady improvements, motivating a re-examination of the statistical formulations and parameterizations in the differential-equation model. Results of this re-examination will be presented.