Comprehensive observation and modeling of earthquake and temperature-related seismic velocity changes in northern Chile with passive image interferometry

We report on earthquake and temperature‐related velocity changes in high‐frequency autocorrelations of ambient noise data from seismic stations of the Integrated Plate Boundary Observatory Chile project in northern Chile. Daily autocorrelation functions are analyzed over a period of 5 years with passive image interferometry. A short‐term velocity drop recovering after several days to weeks is observed for the Mw 7.7 Tocopilla earthquake at most stations. At the two stations PB05 and PATCX, we observe a long‐term velocity decrease recovering over the course of around 2 years. While station PB05 is located in the rupture area of the Tocopilla earthquake, this is not the case for station PATCX. Station PATCX is situated in an area influenced by salt sediment in the vicinity of Salar Grande and presents a superior sensitivity to ground acceleration and periodic surface‐induced changes. Due to this high sensitivity, we observe a velocity response of several regional earthquakes at PATCX, and we can show for the first time a linear relationship between the amplitude of velocity drops and peak ground acceleration for data from a single station. This relationship does not hold true when comparing different stations due to the different sensitivity of the station environments. Furthermore, we observe periodic annual velocity changes at PATCX. Analyzing data at a temporal resolution below 1 day, we are able to identify changes with a period of 24 h, too. The characteristics of the seismic velocity with annual and daily periods indicate an atmospheric origin of the velocity changes that we confirm with a model based on thermally induced stress. This comprehensive model explains the lag time dependence of the temperature‐related seismic velocity changes involving the distribution of temperature fluctuations, the relationship between temperature, stress and velocity change, plus autocorrelation sensitivity kernels. Key Points Superior sensitivity of velocity to stress and acceleration at Salar Grande Linear correlation between velocity decrease and PGA at single station Observation and modeling of velocity changes caused by thermal stresses