Leveraging High-Performance Computing for Enhanced Lava Flow Forecasting Workflow

The integration of lava flow forecasting models with satellite remote sensing techniques marks a significant advancement in quantitative hazard assessment for effusive volcanic eruptions. Within the framework of the DT-Geo project, we are developing a lava flow workflow that harnesses High-Performance Computing (HPC) capabilities, aiming to improve hazard assessment through ensemble-based and data assimilation methods. At the core of the workflow is the VLAVA code, which simulates the lava flow propagation, with temperature-dependent viscosity over a complex topography, erupting from one or more vents. The simulation runs for a given time period (order of one or more days), after which the simulated deposit is compared to the observed lava flow field and, eventually, the observations are assimilated into the model for a further simulation. The measured data include changes of the eruption source parameters and/or the extension and temperature of the lava flow field. These are derived from direct observations on the field or by remote sensing from airborne, drones or satellites (e.g.: Pléiades, EOS-ASTER, SEVIRI, MODIS, VIIRS, Landsat, Sentinel, etc.). Data assimilation is conducted using PDAF, a dedicated software offering various approaches, including ensemble-based Kalman filters, nonlinear filters, and variational methods. The model output provides the potentially impacted area by lava flows, including thickness and temperature distribution, for both a single scenario (utilized for estimating the impact of a lava flow) and an ensemble of weighted scenarios (for generating probabilistic hazard maps). We present the overarching concept of the workflow and share preliminary results obtained for historical eruptions of Mount Etna.