Quantifying the Uncertainty of a Coupled Plume and Tephra Dispersal Model: PLUME‐MOM/HYSPLIT Simulations Applied to Andean Volcanoes

• Published in: Journal of geophysical research. Solid earth 2020-02, Vol.125 (2), p.n/a
• Main Authors: Tadini, A., Roche, O., Samaniego, P., Guillin, A., Azzaoui, N., Gouhier, M., Michieli Vitturi, M., Pardini, F., Eychenne, J., Bernard, B., Hidalgo, S., Le Pennec, J. L.
• Format: Article
• Language: English
• Subjects: Andean volcanoes; Computer applications; Computer simulation; Dispersal; Dispersion; Earth Sciences; Eruptions; Geological hazards; Geophysics; Grain size; Mass; Meteorological data; numerical modeling; Particle size; Sciences of the Universe; Stratigraphy; Tephra; tephra dispersal; tephra fall; Uncertainty; uncertainty quantification; Volcanic activity; Volcanoes; Volcanology
• ISSN: 2169-9313; 2169-9356
• DOI: 10.1029/2019JB018390

Numerical modeling of tephra dispersal and deposition is essential for evaluation of volcanic hazards. Many models consider reasonable physical approximations in order to reduce computational times, but this may introduce a certain degree of uncertainty in the simulation outputs. The important step of uncertainty quantification is dealt in this paper with respect to a coupled version of a plume model (PLUME‐MoM) and a tephra dispersal model (HYSPLIT). The performances of this model are evaluated through simulations of four past eruptions of different magnitudes and styles from three Andean volcanoes, and the uncertainty is quantified by evaluating the differences between modeled and observed data of plume height (at different time steps above the vent) as well as mass loading and grain size at given stratigraphic sections. Different meteorological data sets were also tested and had a sensible influence on the model outputs. Other results highlight that the model tends to underestimate plume heights while overestimating mass loading values, especially for higher‐magnitude eruptions. Moreover, the advective part of HYSPLIT seems to work more efficiently than the diffusive part. Finally, though the coupled PLUME‐MoM/HYSPLIT model generally is less efficient in reproducing deposit grain sizes, we propose that it may be used for hazard map production for higher‐magnitude eruptions (sub‐Plinian or Plinian) for what concern mass loading. Key Points We present an uncertainty quantification for a coupled version of a plume model (PLUME‐MoM) and a tephra dispersal model (HYSPLIT) The model has been tested against field data of four eruptions from Andean volcanoes (in Ecuador and Chile) of different magnitudes/styles The main conclusion of the uncertainty quantification is that the model is best suited for hazard studies of higher‐magnitude eruptions