Efficient application of GPGPU for lava flow hazard mapping

Efficient Lava Flows Simulations with OpenCL: A Preliminary Application for Civil Defence Purposes

“…In accordance to the recent literature (e.g., [6], [19]), in the GPGPU parallel implementation of the model, most of the automaton data (i.e. both the CAcurrent and CAupdated memory areas) was stored in the GPU global memory.…”

“…By adopting a technique well described in [27] and [6], referred to Mt. Etna and by adopting both MPI and GPGPU parallelizations, we here show the preliminary application to a delimited area of the volcano using the OpenCL parallelization of the SCIARA-fv2 model briefly described in Section 2.2.…”

“…Starting from previous related studies [6] referred to CA lava flow modeling with the CUDA framework, this paper presents a preliminary Civil Defence application of a wellknown, reliable and efficient CCA model adopted for lava flow risk assessment, namely the SCIARA-fv2 model [7], in the OpenCL GPGPU environment. Tests performed on different types of GPU hardware, namely a Tesla K40c, a Tesla K20c, a GeForce GTX980 and a Sapphire 280x, and by adopting difference implementation strategies, have shown the validity of this kind of approach.…”

A cellular model for secondary lahars and simulation of cases in the Vascún Valley, Ecuador

“…For both the types of lahar, flowing matter is modeled to be produced by two 20 main mechanisms: (1) a mobilization process of tephra (pyroclastic stratum that could contain some exotic material) is caused by percolation of an adequate water quantity on abundant unconsolidated sediment, and (2) an erosion process depends on the impact of lahar with high kinetic energy on the soil (that could contain more pyroclastic matter than exotic material or vice versa), where such 25 an erosion could be facilitated (but not caused) by rainfall water percolation.…”

“…The previous listed mechanisms were modeled according to the methodolog-60 ical approach of Multicomponent (or Macroscopic) Cellular Automata (MCA) [17][18] [19] for computer simulation of surface flows: water percolation (e.g., model SOIL [17] and model SCAVATU [20]); soil detachments evolving in debris, mud, granular flows (family of models SCIDDICA [21][22] [23]) and deposit formation can be described in terms of viscosity variation by water loss (in sim-65 ilar equational way as viscosity variation of lava flows by cooling in SCIARA model [17][24] [25]). …”