Reduction of influences of the earth's surface fluid loads on GPS site coordinate time series

Conspicuous periodic variations often appear in GPS site coordinate time series. These variations could be influenced by various loads, such as atmospheric, non-tidal ocean, continental water, and snow loads. To eliminate the load influences from the GPS site coordinate time series, we estimated the influences at the GEONET (GPS Earth Observation Network) sites by using meteorological and other loading data sets. Because the GEONET coordinates (F2 solution) suffer scale errors due to a software bug and groundwater variations at the International GNSS Service (IGS) station in Tsukuba, we first corrected these errors. Then, the corrections of loading influences were evaluated for several combinations of the loads. The results show that a combination of atmospheric, non-tidal ocean, continental water, and snow loads can eliminate about 20% of the annual signal in the coordinate time series for both the horizontal and vertical components. They also show that the influence of the non-tidal oceanload is the largest of all the loads. We applied the loading correction to the data of the 1997 Bungo channel slow slip event and showed that the correction can benefit the analysis of sucha non-periodic event