UNIGE document Scientific Article
previous document  unige:40594  next document
add to browser collection
Title

Sensitivity analysis and uncertainty estimation for tephra dispersal models

Authors
Scollo, Simona
Tarantola, Stefano
Coltelli, Mauro
Saltelli, Andrea
Published in Journal of Geophysical Research. 2008, vol. 113, no. B6, p. B06202
Abstract Sensitivity analysis and uncertainty estimation are crucial to the validation and calibration of numerical models. In this paper we present the application of sensitivity analyses, parameter estimations and Monte-Carlo uncertainty analyses on TEPHRA, an advection-diffusion model for the description of particle dispersion and sedimentation from volcanic plumes. The model and the related sensitivity analysis are tested on two sub-plinian eruptions: the 22 July 1998 eruption of Etna volcano (Italy) and the 17 June 1996 eruption of Ruapehu volcano (New Zealand). Sensitivity analyses are key to (1) constrain crucial eruption parameters (e.g., total erupted mass) (2) reduce the number of variables by eliminating non-influential parameters (e.g., particle density) and (3) investigate the interactions among all input parameters (plume height, total grain-size distribution, diffusion coefficient, fall-time threshold and mass-distribution parameter). For the two test cases, we found that the total erupted mass significantly affects the model outputs and, therefore, it can be accurately estimated from field data of the fallout deposit, whereas the particle density can be fixed at its nominal value because it has negligible effects on the model predictions.
Keywords Sensitivity analysisUncertainty estimationTephra dispersal modelsEtnaRuapehu
Identifiers
Full text
Structures
Research group Physical Volcanology and Geological Risk
Citation
(ISO format)
SCOLLO, Simona et al. Sensitivity analysis and uncertainty estimation for tephra dispersal models. In: Journal of Geophysical Research, 2008, vol. 113, n° B6, p. B06202. https://archive-ouverte.unige.ch/unige:40594

191 hits

65 downloads

Update

Deposited on : 2014-09-30

Export document
Format :
Citation style :