Combination procedure for length-of-day time series according to the noisefrequency behavior

Length-of-day (LOD) estimates from seven Global Positioning System (GPS) an d three satellite laser ranging (SLR) analysis centers were combined into a n even-spaced time series for a 27-month period (1996-1998). This time seri es was compared to the multitechnique Earth-orientation-parameter (EOP) com bined solution (C04) derived at the Central Bureau of the International Ear th Rotation Service (IERS/CB). Due to inhomogeneities in the different seri es derived from the various techniques (time, length, quality, and spatial resolution), the concept of a combined solution is justified. The noise beh avior in LOD for different techniques varies with frequency, the data serie s were divided into frequency windows after removing both biases and trends . Different weight factors were assigned in each window, discriminating by technique, and produced one-technique combined solutions. Finally, these on e-technique combined solutions were combined to obtain the final multi-tech nique solution. The LOD combined time series obtained by the present method based on the frequency windows combined series (FWCS) is very close to the IERS C04 solution. It could be useful to generate a new LOD reference time series to be used in the study of high-frequency variations of Earth rotat ion.