Geodetic Observations of Shallow Creep on the Laohushan‐Haiyuan Fault, Northeastern Tibet

Abstract

AbstractWe investigated the spatial distribution of aseismic creep on the Laohushan‐Haiyuan fault using Global Positioning System (GPS) data (1999–2017) and Interferometric Synthetic Aperture Radar (InSAR) data (2003–2010). Comparisons among GPS, InSAR line‐of‐sight (LOS) rates, and leveling show that neither leveling nor GPS vertical velocities can fit the vertical signal mapped into the LOS, implying either complicated vertical crustal deformation in northeastern Tibet and/or complex error structures in the InSAR data. Thus, we combined horizontal GPS with high‐pass filtered InSAR data to produce a continuous LOS rate map crossing the fault. Our geodetic data reveal three creep sections along the fault. Both the restored LOS data and decomposed ascending and descending InSAR data highlight the fact that vertical motion can cause an overestimation of creep rate; we obtained a refined creep rate of 2.5 ± 0.4 mm/a on the Laohushan fault. We further identified a 10 km‐long, ∼3–5 mm/a creep section (∼104.2°E−104.3°E) and a 43 km‐long, ∼1–3 mm/a creep section (∼105.3°E−105.7°E) on the western and eastern Haiyuan fault respectively. Both are located on fault sections that ruptured during the 1920 M∼8 earthquake, suggesting that the 1920 earthquake was able to cross pre‐existing creep sections or that the fault shows heterogeneous relocking after large earthquakes, with creep lasting decades on some parts of the rupture. Fault coupling shows a highly variable rate of slip deficit accumulation along strike, suggesting that coupling might significantly evolve during the period between two large earthquakes.