Multiple taper spectral analysis of earth rotation data

The multiple taper method is optimal for bias control, and yields consistent estimates without section or band averaging. This method was applied to various length-of-day and polar motion data sets from optical, satellite laser ranging (SLR), and very long baseline interferometry (VLBI) techniques. The power spectrum was used to compare length of day data sets samples at different times. The NGS and GSFC reductions of IRIS VLBI data, the IRIS intensive VLBI data (filtered and decimated to 5 day samples), and the CSR SLR data (filtered and decimated to 6 day samples) are consistent with the same model power spectrum. However, the GSFC SLR and CDP VLBI data are not consistent with this model. When the CDP observations are removed from GSFC's data, coherence between the NGS and GSFC data improves and is 0.8 or higher for all frequencies. Lunar occultation length of day must be smoothed. Several data sets were used to construct a composite power spectrum for length of day. This spectrum bends between 1 and 5 cycles per year. The power spectrum of NMC atmospheric data has similar structure. For polar motion, the power spectrum of IRIS VLBI data lies about 15 dB below the ILS power spectrum at high frequencies. The retrograde annual wobble is much smaller in the space geodetic data, suggesting that the optical data is contaminated with an annual systematic error.