Model Validation of Energy Cone Applied to Vesuvius (italy): Insights to Structured Probabilistic Hazard Assessment on Pyroclastic Density Currents

Pyroclastic Density Currents (PDCs) are the most devastating phenomenon related to explosive volcanism. Due to complex nature, their numerical modeling is very challenging and available models range from very simple (e.g., Energy Cone -EC- model) to very sophisticated. The balance between computational demand and output reliability of each model should represent the key point on choosing the ';best' models to use in probabilistic hazard assessment of PDCs. However, scarce model calibration and validation studies have been developed up to now. Here, we present a first approach to understand and estimate the epistemic uncertainty related to numerical modeling of PDCs. The selected model is EC due to two main reasons: 1) It is the simplest model capable to deal with large high-damaging dilute PDCs and 2) The main output variable is the maximum distance (runout) reached by PDCs, which stands as the first parameter to be considered in PDCs hazard assessment and risk mitigation. Aiming at covering the existing natural variability in eruption dynamics (even during an eruption of a given VEI), we have carried out a methodology to sample H and H/L input parameters from reasonably ranged probability density functions, obtaining a coherent set of model runs. We have compared the output values with field observations of actual PDCs at Vesuvius (Italy), in order to account for goodness-of-fit and, also, epistemic uncertainty of the numerical model. More studies on model calibration and validation using diverse techniques (e.g. sensitivity analysis, information criteria, and so forth) applied to several PDC numerical models (e.g. TITAN2D, FLOW3D) must be a short-term target to produce more structured and transparent hazard and risk assessments of PDCs.