Аннотация:Horizontal and vertical station velocities in GPS solutions are affected by uncertainties in translation rate of the Earth center (EC). In conventional reference frames such as ITRF2000 usually applied to constrain a GPS solution, EC is mostly constrained by SLR. Our global GPS solution GPS2006.5 for velocities of 226 stations, observed for at least 3 yrs in 1995-2006, does not use any a priori conventional reference frame, nor any geologic plate motion model. Using GPS observations alone, we solved simultaneously for the current relative plate rotation vectors of nine major plates, EC translation rate, and plate-residual station velocities (horizontal and vertical). EC of ITRF2000 is moving with respect to EC of GPS2006.5 at a rate of 1.1 mm/yr, therefore our estimates of velocities would change within 1 mm/yr if re-evaluated with respect to ITRF2000. According to numerical tests, the translation rate of EC in GPS2006.5 is separated from plate motions with an accuracy better than 0.1 mm/yr. Horizontal and vertical velocities in the regions of the Glacial Isostatic Adjustment (GIA) of the Arctic and Antarctica are evaluated in GPS2006.5 from longer time series and from more stations than in our previous solution [Steblov et al., 2003]. In general, our estimates of the rates of uplift and subsidence are more uniform and vary less within a given region than in published solutions. Our findings for GIA can be summarized as follows. (1) The Arctic. The part of Arctic encompassing the Hudson Bay, Greenland, Fennoscandia, Spitzbergen, and Iceland appears as a single region of uplift with the rate peaking to 5-7 mm/yr over the Hudson Bay and Greenland. Conspicuously, Arctic Siberia is not rising, as evidenced by a representative set of stations. This fact is a strong confirmation that Siberia was not significantly glaciated during Last Glacial Maximum (LGM), in agreement with RSL observations as reflected in ICE-4G deglaciation model [Peltier, 2004]. The rate of uplift around the Hudson Bay and over Greenland revealed by GPS is in good agreement with the ICE-4G (but not ICE-5GP). In contrast, GPS plate-residual horizontal velocities in North America disagree with ICE-4G. (2) Antarctica. GPS velocities in Antarctica are only estimated from stations placed on the margin of the continent. All the stations except two are rising at a rate of 1-2 mm/yr, which is much smaller than the uplift rate found in the Arctic. VESL is falling at a rate less than 1 mm/yr and OHIG is rising at 6 mm/yr. Plate-residual horizontal velocities are 1-2 mm/yr, again excluding OHIG (5 mm/yr). 5 stations of totally 7 are moving towards interior of Antarctica, opposite to typical theoretical predictions of GIA for horizontal velocities [Milne et al., 2001; Peltier, 2002]. We expect that our estimates will help improve understanding of GIA for the regions examined here, since competing models differ significantly in the predicted uplift rate and horizontal velocities.