Fault slip vs. sea sea surface tsunami source inversion. Two approaches to tsunami source inversion: (a) Slip on a fault is assumed and translated to surface deformation, requiring several assumptions about fault geometry, material properties, and dislocation modeling for the computation of Green’s functions (GFs). (b) The approach applied here computes GFs directly for elementary waves at the sea surface and inverts for the tsunami source directly without assumptions about the fault and seafloor deformation. (From Dettmer et al, 2016, Tsunami source uncertainty estimation: The 2011 Japan tsunami. JGR-Solid Earth, 121(6), 4483–4505.)

Nature of Project:

Earthquake and tsunami source imaging

Essential Background:

PHYS 3070 (Physics of the Earth); EMSC 8023 (Advanced Data Sciences).


Earthquakes and tsunamis are rapid-onset hazards that can have massive impacts on societies in tectonically active regions and coastal communities. Earthquakes generate strong ground shaking that can damage buildings and infrastructure, causing death and injuries through building collapse and mass failure, while tsunami inundation can obliterate entire communities in low-lying coastal zones. In order to help communities build resilience to earthquake and tsunami disasters, we need to improve our understanding of earthquakes so we can answer questions like: How big and how frequently will earthquakes and tsunamis occur? What factors influence the effectiveness with which earthquakes generate strong ground motion and tsunamis? How does one major earthquake trigger another, and how are earthquake and volcanic activity linked?

One of the most important tools we have for studying earthquake behaviour are observations of the spread of rupture over a fault during the occurrence of an earthquake. Estimates of spatio-temporal evolution of earthquake rupture are important because they provide:

  • Some of the best constraints we have on the physics of earthquake rupture.
  • Fundamental information on the generation of strong ground motion and tsunamis.
  • Key parameters - such as seismogenic zone extent and maximum slip - that are need for recurrence and hazard assessment.

Earthquake and tsunami source imaging is a key tool of earthquake and tsunami science, involving data as diverse as personal observations of ground shaking and mass movement, to precise instrumental observations of seismic and tsunami waveforms, to satellite observations of co-seismic ground displacement.

Possible Future Research Avenues:
Precise estimation of earthquake rupture is a very active research field, with new algorithms and data types – not to mention new events with exotic rupture patterns – appearing frequently. The challenge is to achieve high precision in estimates of rupture, while being able to assess just what this precision is. There are several ways to undertake earthquake and tsunami source imaging:

Finite Fault Inversion (FFI) using seismic or tsunami waveform data. FFI typically uses a very straightforward least squares approach to match observed earthquake or tsunami waveforms with those computed for a grid of point sources covering the fault surface. A variety of seismic waves – body, surface and W-phase – and tsunami waveforms can be used, with each wave type having different sensitivity to different facets of the spatio-temporal evolution of fault rupture, and each presenting its own challenge to high-precision computation. More advanced inversion techniques are required for rigorous uncertainty assessment and full treatment of nonlinearity, the effective implementation of which is an area of active research. Projects in FFI could range from imaging of recent earthquakes using the traditional least-squares approach, to implementations of more novel approaches to source inversion.

Back projection or Reverse Time Imaging (BP/RTI) of seismic and tsunami waveforms makes use of the principle of time-reversal symmetry to refocus time-reversed waveforms. - either seismic or tsunami – to image the earthquake/tsunami source. The data used range from sea level measurements from seafloor pressure sensors in the deep ocean, to regional seismic networks or tightly coupled arrays of seismic sensors, and the approaches range from simple “delay and sum” techniques to full numerical simulations of the reverse-time wavefield. While less precise than source inversion, BP/TRI often proves very effective in estimating the extent and spread of rupture propagation, and in identifying the parts of the fault where most energy was released as seismic waves. It is often fast enough to be considered for use in early warning systems.
Projects in BP/RTI could use techniques developed at RSES for precise modelling of deep ocean tsunamis for reverse time imaging of recent tsunamis, or consider the use of back projection of Australian seismic arrays to constrain rupture properties of regional earthquakes.

Joint inversion of multiple data types. Over the past decade, the occurrence of large earthquakes have generated rich datasets from a range of observation platforms far more diverse than was the case even as recently as the early 2000s. In addition to the proliferation of high-quality, broadband seismic networks, data from continuous GPS networks, sea level buoys and seafloor pressure gauges, satellite data of unprecedented quality and coverage is now available, often only days after the earthquake. Combining these data types in an inversion that will optimally resolve the spatio-temporal evolution of slip on the fault is a challenging research problem that as yet has no definitive solution. However, with their potential to dramatically increase the spatial resolution of earthquake rupture models, mixed-data-type inversions will potentially yield invaluable new insight into rupture dynamics and tsunami generation.