1 Purpose
The Alert Level module within Exupéry VFRS allows to automatically compute an alert level of the volcano. It is based on Bayesian Belief Networks (BBN) and fetches as well as provides data from and for the database. More details on the general purpose can be found here.
However, there is no way to replace the volcanologist himself. The automatic alert level gives the expert just a hint in which direction it could be. The following example might clarify the idea:
The volcanologist gives the alert level 1 (the lowest one) based on his experience and on all the observations which he accesses via the GIS. The automatic alert level system provides alert level 3 (the highest one). In this case the volcanologist is encouraged to double check his decision. The alert level module provides the possibility for the volcanologist to download the alert level network file (which includes the observed parameters) load it into a BBN software and inspect the reason (probabilities and observations) why the automatic system provides such a high alert level.
(The used bayesian network is a modified version of a bayesian network originally provided by Willy Aspinall.)
The software installation is treated in the following.
Literature:
[1] W.P. Aspinall, G. Woo, B. Voight, and P.J. Baxter. Evidence-based volcanology: Application to eruption crises. JVGR, 128 (13), 273---285, 2003.
[2] T. Hincks, Probabalistic volcanic hazard and risk assessment , PhD Thesis, University of Bristol, 2006. [link]