We use GPS data from 45 sites across the Tibetan Plateau surveyed between 1991 and 2001 to study the distribution of strain in that part of the India-Eurasia collision zone. The plateau is cut by a few major, rapidly slipping strike-slip fault zones, with broadly distributed strain between those zones. The GPS velocities can be fit well by a simple deforming block model that combines uniform strain with the motion of blocks separated by the major known fault zones within the plateau. The boundaries of the four blocks in this model correspond to the major, rapidly slipping faults. A rigid block model with a smaller number of blocks fits the data poorly and can be rejected. We estimate that 5.9 +/- 0.7 mm/yr of extension in the N69degreesW direction occurs on the Yadong-Gulu rift south of the plateau; localized extension may extend as far north as the Nyainqentanglha Range. We find 7.4 +/- 0.7 mm/yr of right-lateral slip on the Karakorum-Jiali fault zone, significantly slower than that previously estimated from offset geologic features. Our deforming block model and a two-dimensional single-fault screw dislocation model give lower and upper bounds of 4.4 +/- 1.1 and 10.3 +/- 0.4 mm/yr on the slip rate on the Kunlun fault, respectively, comparable, in its highest, to the long-term slip rate observed geologically. The distributed deformation is surprisingly uniform over the Tibetan Plateau, Qaidam Basin, and Qilian Shan and approximates a combination of pure shear and uniaxial contraction with the same axes of maximum contraction, similar toN32degreesE. This strain field is a combination of shortening and extension that produces little net dilatation (1% area loss per million years) at present. The broadly distributed strain most likely represents slip on many faults, each with a relatively low slip rate. Distributed conjugate strike-slip faulting is the most plausible mechanism to produce the observed strains, as supported by the record of medium to large earthquakes within the plateau. The distributed deformation is just as important in the accommodation of the total India-Eurasia convergence as is slip on the major faults. The eastward motion of central Tibet is as large as 50% of the convergence rate between India and Eurasia. However, this eastward motion is not motion of a relatively undeforming block bounded by rapidly moving strike-slip faults, as suggested by past extrusion models. Instead, much of this eastward extrusion is driven by the internal extension of the plateau. Our inference is in agreement with the model of F. Shen et al. [2001] that suggested east-west extension, north-south contraction, and eastward plateau growth dominate the present tectonics of the Tibetan Plateau.
