Data di Pubblicazione:
2019
Abstract:
A fully-coupled from surface to groundwater hydrological model is being developed
based on the Extended Cellular Automata formalism (XCA) [1], which proves to be
very suitable for high performance computing. In this note, the module related to
three-dimensional saturated flow in porous media is presented by using the OpenCAL
library [2], allowing to exploit multi-core CPUs and many-core devices like GPUs. The
XCA-based model is evaluated in terms of both accuracy and precision of modeling
results and computational performance, using two- and three-dimensional test cases at
different resolutions (from parcel to catchment scale), simulating pumping from one or
more wells, river-groundwater interactions and varying soil hydraulic properties. Model
accuracy is compared with analytic solutions, when available, and with the widely used
MODFLOW model [3], while the computational performance is evaluated using both
CPU and GPU devices. Overall, the XCA-based model proves to be accurate and,
mainly, computationally very efficient thanks to the many options and tools available
with the OpenCAL library.
Tipologia CRIS:
04.01 Contributo in Atti di convegno
Keywords:
High-performance Computing; Extended Cellular Automata; Computational Fluid Dynamics
Elenco autori:
Giordano, Andrea
Link alla scheda completa: