Abstract by Joshua Fullwood
Jared Whitehead, Ron Harris
Statistical Management of Geophysical Uncertainty in Monte Carlo Tsunami Reconstruction
Our ultimate goal is to help save lives from tsunami hazards by using a Monte Carlo model to reconstruct historical tsunami events with the GeoClaw solver as a black box. This reconstruction relies on comparison between the black box output and historical eye-witness observations of tsunamis like wave height and inundation length. Uncertainty about the geophysical environment is a source of error that must be understood and managed for the statistical results to be considered reliable. A chief problem is the lack of high resolution near-shore bathymetry and onshore topography. This coarse data causes significant mismatch between the model output and the actual observations. We present a statistical solution, generating a probability distribution of on-shore behavior using data from the 2011 tsunami and earthquake in Tohoku, Japan. This allows extrapolation of probable onshore behavior from the GeoClaw output as required.