A quantitative approach for evaluating lava flow simulation reliability: LavaSIM code applied to the 2001 Etna eruption

Abstract

This work developed a quantitative approach for evaluating the reliability of lava flow simulation codes. In particular, it applied the LavaSIM code to simulate the main lava flow emplaced on the south flank of Mount Etna (Italy) between 18 July and 9 August 2001 which represents an ideal test case for validating numerical codes. LavaSIM is the only full 3-D model and is thus able to account for the vertical variation of lava properties such as temperature, viscosity, velocity, and liquidus or solidus state. It presents the most complete description of the lava cooling, and its greatest peculiarity is the potential to discriminate between cells filled by liquid or solid lava. Thirteen simulation tests were performed varying the main input parameters, and they were checked thanks to the availability of syneruption maps, defining the lava flow planar expansion throughout its whole emplacement. Two parameters were adopted for quantitatively evaluate the agreement between real and simulated flows: the percent length ratio (PLR), here defined, and the fitness function (e1). Their joint analysis allowed checking both the simulated lateral spreading, through e1, and the flow lengthening, through PLR. The simulated flows follow a path very similar to the observed one, giving a good fitting of the lateral spreading, though the simulations are, after the second day, normally longer and thinner. The temporal evolution of the three-dimensional distribution of liquid lava and crust was also analyzed. Finally, the analysis presented here demonstrated the great capability of the LavaSIM simulation code.