CORRESPONDENCE OF EOP AND GEOMAGNETIC FIELD

Connection of EOP with centred and eccentric geomagnetic dipole fields is described (with the use of Gaussian coefficients) for the epoch 1880-2000. The main tools of statistical investigation carried out were Laplace-type robust estimation (in case of ∆LOD) and the non-linear regression analysis (for PM). On the basis of computations carried out it can be concluded that the statistical comparison of temporal variation of earth magnetic and EOP data shows significant correlation. 1. DECADAL VARIATIONS OF LOD AND GEOMAGNETIC DIPOLE FIELD Dependence of LOD from processes of the liquid outer core is especially strong in case of periods between 20 and 60 years. This phenomenon determined first of all by core mantle interactions. These interactions can be of different kind. Bloxham (2000) and Gubbins et al. (2003) describe the sensitivity of the axial geomagnetic dipole to thermal interactions at the CMB. Mound and Buffett (2006) detected gravitational oscillations in the LOD. Yukutake (1972) and Yokohama and Yukutake (1992) explain the connection between sixty-year variation in length of day and magnetic field with the electromagnetic core-mantle coupling. Extension of the range of the link between variations of geomagnetic field and LOD in direction of shorter than 20 year periods is problematic because the worldwide geomagnetic data are available recently with a time step of 5 years. More principal difficulty is related to the strong influence of the atmospheric angular momentum on LOD in case of annual/sub-annual periods (Abarca del Rio, 2003), what hides the possibly existing core-mantle interactions. In case of longer, secular, periods one meet the problem of sparse LOD data and indefinability of dating both the magnetic (archeomagnetic) and LOD data. For the aims of present investigation in case of decadal variations the geomagnetic data were obtained from IAGA global spherical expansion database for the time-interval 1880-2000 with a five-year step. Length of day variation data for the same time-interval (and digitalization) are given in form of mean annual values derived from long-term data set of IERS. 2. STATISTICAL BACKGROUND For the study of correlation of geomagnetic and EOP data the robust estimation technique were used together with well-known non-linear regression analysis. The robust estimation is a generalization of least-squares method which reduces the influence of outliers and allows optimal solution in case of different error distribution. According to authors numerous mathematical tests for correlation of LOD and geomagnetic dipole moment the best fit is in case of use of Laplace robust modelling (Laplace type modelling means: the central part of the distribution curve is normal while the wings are linear). In case of correlation of PM and geomagnetic dipole moment the non-linear regression analysis of degree n=6 was in use. In Table 1 the correlation coefficients both for three components of central geomagnetic dipole and for the four components of eccentric dipole field are collected together. The connection of X and Y components of PM are shown and described by Figs. 1a and 1b.