Development of Robust and Efficient Methods for Hydraulic/Hydrologic/Environmental Risk Numerical Simulation

Nowadays, the great power of modern computers combined with well-designed numerical models allows to develop computational models able to deal with simulations of several coupled phenomena over detailed complex topography. An efficient and properly calibrated computational model represents a useful tool to provide insight into the catchment dynamics at hydrological and geomorphological levels. In addition, it allows to develop detailed risk management and conservation plans. The challenge of finding a compromise between computational time and level of accuracy and robustness has traditionally expanded the use of depth averaged models (2D or even cross section averaged models (1D) rather than full three-dimensional models for flood simulation. This work presents examples of GPU accelerated 2D shallow-water models for the simulation of flood events, water quality evolution problems and urban flow over non-erodible and erodible bed in real time. In particular, a well-tested and robust explicit first-order finite volume scheme is used to control the numerical instabilities that are likely to appear when used in complex topography. The model is applied to reproduce real events in a reach of the Ebro River (Spain) and a tailing dam in Brumadinho (Brazil) in order to compare simulation results with field data in a large domain and long flood duration allowing an analysis of the performance and speed-up achieved by different GPU devices. The high values of fit between observed and simulated results as well as the computational times achieved are encouraging to propose the use of the model as forecasting system.