This data set contains velocity and strain rate fields over the northeast Tibetan Plateau, which are derived from Sentinel-1A and -1B synthetic aperture radar satellite data (SAR) and stored in GeoTIFF (.tif) or NETCDF (.grd) formats.
The velocities in the line-of-sights (LOS) of the satellites were processed at ~100 m resolution from time series in ~250km x 250km frames. The data set consists of velocities from 10 frames in ascending tracks and 13 frames in descending tracks of the satellites' orbits. The spatial extent of the velocities spans 96E-108E and 32N-43N, covering an area of 660,000 km^2. The temporal coverages of the data span from October 2014 to December 2019 across 65-110 acquisition epochs. The uncertainties of the velocities average to <1 mm/yr. The time series are inverted from fully-connected networks of short-temporal-baseline interferograms which are generated from interfering and unwrapping pairs of SAR imagery. The velocities represent the average velocity through the displacement time series.

The LOS velocities were decomposed into east and vertical velocities which are also archived with associated uncertainties. These Cartesian fields cover the overlapping areas between ascending and descending tracks and total 440,000 km^2. By combining the horizontal gradients of the filtered east velocities and interpolated north velocities from Global Navigational Satellite System, we derive second invariant, maximum shear, and dilatation strain rate fields for the same area with 1 km sampling intervals.
These strain rate fields highlight creeping sections and strain concentration on faults and fault junctions. The velocity fields reveal fault kinematics in terms of slip rates and partitioning. The vertical velocities also show non-tectonic signals such as subsidence related to permafrost melting, groundwater extraction, and reservoir loading, as well uplift from blocked drainages.

The data are collected and processed by Qi Ou with the automatic processing tools developed by Milan Lazecky. Velocity and strain rate fields were interpreted by all authors. By default, interferograms were generated from each epoch to six consecutive epochs and between acquisition pairs with six-month and nine-month temporal baselines. Interferograms with the unwrapping error were removed from the network and all networks were continuous and fully connected.