Novel statistical emulator construction for volcanic ash transport model Ash3d with physically motivated measures

Abstract

Statistical emulators are a key tool for rapidly producing probabilistic hazard analysis of geophysical processes. Given output data computed for a relatively small number of parameter inputs, an emulator interpolates the data, providing the expected value of the output at untried inputs and an estimate of error at that point. In this work, we propose to fit Gaussian Process emulators to the output from a volcanic ash transport model, Ash3d. Our goal is to predict the simulated volcanic ash thickness from Ash3d at a location of interest using the emulator. Our approach is motivated by two challenges to fitting emulators—characterizing the input wind field and interactions between that wind field and variable grain sizes. We resolve these challenges by using physical knowledge on tephra dispersal. We propose new physically motivated variables as inputs and use normalized output as the response for fitting the emulator. Subsetting based on the initial conditions is also critical in our emulator construction. Simulation studies characterize the accuracy and efficiency of our emulator construction and also reveal its current limitations. Our work represents the first emulator construction for volcanic ash transport models with considerations of the simulated physical process.

Calibration strategies of PDC kinetic energy models and their application to the construction of hazard maps

“…Inter-comparison of models is in fact a critical step in the validation of numerical tools, as discussed by Esposti Ongaro et al (2020a), being particularly relevant when these tools are used to define measures of volcanic risk mitigation. We stress that there are similar difficulties in any numerical model when it is adopted to describe a physical phenomenon characterized by significant uncertainty (Scollo et al, 2008;Worni et al, 2012;Biass et al, 2016;de' Michieli Vitturi and Tarquini, 2018;Bevilacqua et al, 2019;Yang et al, 2020), which appeals to the development of strategies to address this issue.…”

Toward Next‐Generation Lava Flow Forecasting: Development of a Fast, Physics‐Based Lava Propagation Model

“…In any case, the parameter space will be multidimensional, requiring many ensemble members to produce a robust forecast. Integration of such a parameter distribution could proceed simply by direct Monte Carlo methods such as random sampling or Latin Hypercube sampling (e.g., Bevilacqua, Patra, et al., 2019) or by surrogate methods such as Gaussian Stochasitc Process (GaSP) emulation (e.g., Yang et al., 2020) or Polynomial Chaos Expansion (PCE) methods (e.g., Bursik et al., 2012) if a high‐dimensional parameter space must be used or if a small set of runs is required for timeliness. See Poland and Anderson (2020) for a more complete review of ensemble forecasting considerations in volcanology.…”

Assessing minimum pyroclastic density current mass to impact critical infrastructures: example from Aso caldera (Japan)

“…This enables us to utilize a very fast model inversion approach in the uncertainty quantification process. We note that we are not using "reduced" models (i.e., statistical surrogates, e.g., Rutarindwa et al, 2019;Yang et al, 2020). Further details on the physical equations we adopted as well as the mathematical expression of the analytical solutions, can be found in the Supplement S1.…”