norme di preparazione dei manoscritti
Transcript
norme di preparazione dei manoscritti
Lava flow simulations by MAGFLOW model on CUDA architecture Bilotta G. (1, 2), Cappello* A. (1, 2), Del Negro C. (1), Gallo** G. (2), Ganci G. (1), (3) (1, 2) (1) Hérault A. , Rustico E. , Vicari A. (1 ) Istituto Nazionale di Geofisica e Vulcanologia (INGV), Piazza Roma 2, 95123 Catania, Italy E-mail: [email protected], [email protected], [email protected], [email protected] (2) Dipartimento di Matematica e Informatica, Università di Catania, Viale Andrea Doria, 95125 Catania, Italy E-mail: [email protected], [email protected], [email protected] (3) Conservatoire National des Art et Métiers, Département de Mathématiques, 292 rue Saint-Martin, 77141 Paris, France E-mail: [email protected] * Corresponding author **Conference speaker Abstract Developing physical-mathematical models capable to describe the spatial and temporal evolution of volcanic flows is essential to support impact mitigation action and land planning, in combination with laboratory and field observations. Physical models of lava flows must take into consideration the non-linear, temperature-dependent rheology, the variation in space and time of the lava parameters, the irregularity of natural topographies. The more physically correct a model, the more computationally intensive it is, a condition that can hinder the applicability of the model for short-term forecasting of lava emplacement during an eruption, for which it is essential that scenario simulations be completed in very short times compared to the actual phenomenon evolution. That challenge has inspired the INGVCT to develop MAGFLOW model (Vicari et al., 2007) that includes a physical description of lava flow propagation, allowing us to effectively simulate the time of advancing and the maximum length of the lava flow. MAGFLOW is based on a cellular automata structure in which the evolution function is a steady state solution of the Navier-Stokes equations and heat transfer (due to radiative losses), and solidification effects are modeled via a temperature dependent viscosity. It has been successfully employed at Etna volcano to reproduce past events with well-determined characteristics and to predict lava flows in real time during the 2004, 2006 and 2008 eruptions. More recently, it has been applied to more sophisticated issues including prospective hazard mitigation (Scifoni et al., 2010) and for compiling a new lava flow invasion hazard map at Etna volcano (Cappello et al., 2010). Although the original version of MAGFLOW is intended for serial execution on standard CPUs, the cellular automaton paradigm displays a very high degree of parallelism that makes it suitable for implementation on parallel computing hardware. In particular, we chose to implement the MAGFLOW model using CUDA (Compute Unified Device Architecture), an architecture provided by NVIDIA for the deployment of their last generations of Graphic Processing Units (GPUs) as high-performance computing hardware. The conversion of the serial CPU code leads to a 50x speedup in execution on the last generation of CUDA cards, with negligible loss in precision. This means that MAGFLOW on GPUs provides the result of a simulation lasting seven days in a couple of minutes. The prompt control of the MAGFLOW simulator on CUDA architecture has been gathered in the forecasting module of LAV@HAZARD (Vicari et al., 2011), a newly-developed information system dedicated to promote mitigation actions thanks to the high degree of interactivity, easily readable maps (based on Google Maps Application Programming Interface) and a fast way to explore alternative scenarios (Fig. 1 ). Figure 1 – Screenshot of the Forecasting module of LAV@HAZARD. A simulation performed by MAGFLOW model on CUDA architecture with the main input parameters is displayed. Colors are associated to different thickness or to temperature of the flow. References Cappello A., A. Vicari and C. Del Negro (2010). Assessment and Modeling of Lava Flow Hazard on Mt Etna Volcano. Bollettino di Geofisica Teorica e Applicata, in press. Scifoni S., M. Coltelli, M. Marsella, C. Proietti, Q. Napoleoni, A. Vicari and C. Del Negro (2010). Mitigation of lava flow invasion hazard through optimized barrier configuration aided by numerical simulation: The case of the 2001 Etna eruption. Journal of Volcanology and Geothermal Research 192, 16–26. Vicari A., A. Hérault, C. Del Negro, M. Coltelli, M. Marsella and C. Proietti (2007). Modeling of the 2001 lava flow at Etna volcano by a cellular automata approach. Environmental Modelling & Software, 22, 1465-1471. Vicari A., G. Bilotta, S. Bonfiglio, A. Cappello, G. Ganci, A. Hérault, E. Rustico, G. Gallo and C. Del Negro (2011). LAV@HAZARD: A Web-Gis interface for volcanic hazard assessment, Special Issue of Annals of Geophysics.