Reconciling Geodetic Strain and Seismicity Rate with Frequency-Magnitude Relation of the Largest Earthquakes

Abstract

The aim of this thesis is to study how moment buildup rate on faults can be reconciled with moment release rate. We concentrate first on the Himalaya region and go on to look at faults worldwide. We first justify the extrapolation of GPS data in the Himalayan region over the approximate timescale of an earthquake cycle. To do this we show that GPS strain rates correlate with seismicity rates, and that the principal directions of strain found from GPS data are similar to those from earthquake moment tensors, showing that GPS data has been consistent at the timescale of earthquake strain-rate build-up, roughly 100-1000 years. We next use geodetic data to show that the Main Himalayan Thrust (MHT) is locked from the surface to roughly 100 km north along its entire length, with no creeping patches. We also find the long-term slip rate on the fault, and these values agree with values from geomorphic studies, showing that here the tectonic regime has been stable with time, and most of the deformation is elastic. However, we also find a correspondence between the pattern of uplift rate predicted from the model and the topography, suggesting that a small amount of permanent deformation (10%) may occur, and again suggesting that the pattern of coupling has been stable with time. We find the moment build-up rate on the MHT to be 15.1±1.0x1019 Nm/yr and compare this rate with the rate of moment release estimated from large earthquakes that have occurred on this fault in the past 1000 years. We use the conservation of moment principal to model the most likely maximum magnitude earthquake that needs to occur to balance the moment budget, and find that we need an earthquake of magnitude 9 or more with a recurrence time of roughly 800 years. We extend this analysis to faults with no GPS data, and no long record of large earthquakes, by developing a method to find the expected maximum magnitude earthquake on faults assuming conservation of moment, and that the earthquakes follow the Gutenberg-Richter law. Our results compare well with historical catalogs where they are available.