It is widely believed that groups of hot spots in different regions of the world are in relative motion at rates of 10 to 30 mm a−1 or more. Here we present a new method for analyzing geologically current motion between groups of hot spots beneath different plates. In an inversion of 56 globally distributed, equally weighted trends of hot spot tracks, the dispersion is dominated by differences in trend between different plates rather than differences within plates. Nonetheless the rate of hot spot motion perpendicular to the direction of absolute plate motion, vperp, differs significantly from zero for only 3 of 10 plates and then by merely 0.3 to 1.4 mm a−1. The global mean upper bound on |vperp| is 3.2 ± 2.7 mm a−1. Therefore, hot spots move slowly and can be used to define a global reference frame for plate motions.

The Qaidam basin is the largest topographic depression inside the Tibetan Plateau and it is a key factor to understanding the Cenozoic evolution of the northern Tibetan Plateau. Paleomagnetic data was obtained from the middle to late Eocene Xiaganchaigou Formation and the early to middle Miocene Xiayoushashan Formation from seven localities. The paleomagnetic results indicate that the Qaidam basin has not undergone obvious basin-scale vertical axis rotation with respect to the Eurasia Plate since the Eocene. Local clockwise rotation took place only at a few special locations along the northern margin of the Qaidam basin. The uniform paleomagnetic results at different localities support that the Qaidam basin is a relatively rigid block. Regional paleomagnetic and geodetic observations also suggest that crust south of the Kunlun fault moves eastward faster than crust north of the Kunlun fault.

Uncertainties in trends of hot spot tracks are investigated using a relationship between trend uncertainty and the mapview dimensions of a hot spot track. Prior estimates of Δt (the time span averaged in estimating the trend of a hot spot track), combined with an observed average track width of σwidth = 33 km, indicate that uncertainties in track trend are larger than estimated before, especially for hot spot tracks on slow‐moving lithosphere. Measured values of σwidth of different hot spot tracks differ insignificantly from one another. Track widths show no significant differences between oceanic and continental tracks and between tracks of deep plumes and tracks of shallow plumes. We find that motion between groups of hot spots on different plates is slow. Nominal speeds vary from 0 to 6 mm/a with a lower bound of zero and upper bounds of 4 to 13 mm/a for the eight best constrained hot spot groups.

The Cenozoic evolution of the Qaidam basin, especially its paleostress field, can provide a better understanding of the dynamistic process of the northern Tibetan Plateau. Under certain conditions, Anisotropy of Magnetic Susceptibility (AMS) holds great potential for investigating early tectonic events, even where macroscopic and microscopic evidence of deformation is invisible. A basin-scale AMS study of the middle to late Eocene Xiaganchaigou Formation and the early to middle Miocene Xiayoushashan Formation from seven locations was conducted, covering most outcrops of these two formations within the Qaidam basin. In the western Qaidam basin, principal stress directions inferred from AMS ellipsoids consist with those inferred from fold axial traces, while at Eboliang and in the northern Qaidam basin, most principal stress directions reflected by AMS ellipsoids are different from those reflected by fold axial traces. Two epochs of compressive strain have been identified: an early N-S strain no later than Oligocene and a late NE-SW strain since Miocene. The early N-S compression is more intense in the northern Qaidam basin than that in the western Qaidam basin, while the late NE-SW compression, which dominates the modern NW-SE trending fold axial traces, is more intense in the western Qaidam basin than that in the northern Qaidam basin. The stress transfer provides a reasonable explanation for the southeastward migration of the deposition center in the Qaidam basin during Cenozoic. Moreover, the appearance of E-component compression may be in close relationship with the beginning of the left-lateral strike-slip Kunlun Fault or the eastward channel flow to the south of the Kunlun Fault.

The Global Moving Hotspot Reference Frame (GMHRF) has been claimed to fit hot spot tracks better than the fixed hot spot approximation mainly because the GMHRF predicts ≈1,000 km southward motion through the mantle of the Hawaiian mantle plume over the past 80 Ma. As the GMHRF is determined by starting at present and calculating backward in time, it should be most accurate and reliable for the recent geologic past. Here we compare the fit of the GMHRF and of fixed hot spots to the observed trends of young tracks of hot spots. Surprisingly, we find that the GMHRF fits the data significantly worse (p = 0.005) than does the fixed hot spot approximation. Thus, either plume conduits are not passively advected with the mantle flow calculated for the GMHRF or Earth's actual mantle velocity field differs substantially from that calculated for the GMHRF.