Inversion of tsunami characteristics:Estimation of transient flow depth and speed with quantified uncertainties

Tuesday, 16 December 2014: 5:45 PM
Hui Tang, Robert Weiss, Heng Xiao and Jianxun Wang, Virginia Polytech State University, Blacksburg, VA, United States
Tsunami deposits are recordings of tsunami events, containing the information about the flow conditions during events. Deciphering quantitative information from the deposits is especially important for paleo events where deposits are the only left-over physical evidence. Inversion of the flow conditions has been attempted in the past. One summarizing conclusion from the different inversion models is that it is a difficult endeavor, and the physical meaning of the inverted quantities depends on the physical assumptions that are applied. The aim of our study is to relate the time-varying characteristics of tsunamis with the deposits, and quantify the error and uncertainty that go with it. For this, we combine TsuSpeedv1 for the deposition with the Ensemble Kalman Filter (EnkF) method to study the deposition of an idealized deposit by one tsunami wave. In our modeling, we assume that information from the idealized deposit at different depths within the deposits can be used as observations, and the coupling between TsuSpeedv1 and EnkF allows us to correct for the different flow conditions causing deposition as the tsunami travels over a certain area. Applying an idealized deposit enables us to study the uncertainty and error that accompanies the inversion process by, for example, varying the number of unknown variables that we aim to invert, or how many observations are available, among others. Our tentative results indicate that sampling methods and sampling frequencies of tsunami deposit influence not only the magnitude of the inverted variables, but also their error and uncertainty. An interesting result of our technique is that a larger number of samples from a given tsunami deposits does not automatically mean that the inversion of, for example, flow speed and flow depth is more robust with smaller error and decreased uncertainty. The same also holds for the number of measured grain-size classes in the deposits. From a more general viewpoint, these two examples indicate that the gap between results from modeling and fieldwork needs to be narrowed or closed because only together more robust inversion can be achieved. In this sense, researchers focusing on fieldwork must understand more about modeling, and more theoretical researchers must better comprehend the limitations and constraints of fieldwork.