3D surface velocities and strain rate fields for the central-eastern Altyn Tagh fault (NW Tibet)
Description
This dataset includes surface velocities (at 1 km resolution) and strain rate fields for a total area of ~ 600,000 km2 (~ 1,300 km × 450 km, Longitude 84°E-100°E, Latitude 36°N-42°N) around the central and eastern segment of the Altyn Tagh fault using Sentinel-1 InSAR rate maps. This dataset can be used to study interseismic strain accumulation and its termination on the central-eastern Altyn Tagh fault and surface deformation of the northwestern Tibet. We used the LiCSAR system (Lazeckỳ et al., 2020) to derive interferogram networks. We processed 11 LiCSAR frames on 7 ascending tracks and 9 LiCSAR frames on 6 descending tracks. For each frame, we used ~170 acquisition epochs between October 2014 and July 2022. To reduce the impact of phase biases and nontectonic seasonal signals, we combine both short temporal (< 4 months) and 1-year to 7-year long summer-to-summer baseline interferograms in the network, which generates an average of nearly 2000 interferograms in each LiCSAR frame. We carried out time-series analysis using LiCSBAS (Morishita et al., 2020), during which the tropospheric phases for each epoch were removed based on the Generic Atmospheric Correction Online Service (GACOS) (Yu et al., 2018). As the remaining interferograms may still have unwrapping errors, we chose to reduce the impact of such errors by nullifying (removing the values) all the displacements of interferograms associated with an unclosed loop in the time series for each pixel. After obtaining displacement time series of each pixel in a frame, the average linear velocities were calculated based on the standard approach in LiCSBAS. Final line-of-sight (LOS) velocity uncertainties are calculated from the standard deviation (STD) of 100 velocities based on resampled datasets of displacement time series using the bootstrap method. We masked relatively unreliable pixels using several noise evaluation indices such as the average coherence and number of network gaps for each pixel (Morishita et al., 2020). We derived reference frame adjustment parameters for each frame using VELMAP (Wang & Wright, 2012; Wang et al., 2019) to tie all the LOS velocity maps to the same Eurasian reference system as the GNSS data. We decomposed the final LOS velocity field accounting for the local radar incidence and azimuth angle. We inverted for the east and vertical velocity , as well as their associated uncertainties, for each pixel based on a method (Watson et al., 2022) in which the solution of north velocities from the VELMAP inversion was projected in local LOS direction and subtracted from the original LOS velocities first. We calculated our horizontal strain rates follow the method from Ou et al. (2022), in which the horizontal strain rates are calculated based on the median filtered east velocities at the InSAR resolution and interpolated north velocities from GNSS (the north velocities from VELMAP in this study).