NA - Large-Scale 3D Transient Hydraulic Tomography Using Parallel Computers

The capability of characterizing heterogeneous hydraulic parameters in the subsurface is essential for monitoring and predicting groundwater resources. Investigating such heterogeneity by the technique of hydraulic tomograph in basin-scale presents a for midable challenge. The size of computation is potentially huge, both due to a 3D domain involving millions of degrees of freedom, and due to a transient parabolic partial differential equation (PDE) that needs to be solved repeatedly, using gradually upda ted coefficients. Because of the numerical complexity and required PDE software, the research group led by Prof. Jim Yeh at Dept. Hydrology and Water Resource, Univ. Arizona wishes to join force with Simula Research Laboratory via Dr. Xing Cai, who speci alizes in designing efficient numerical methods for PDEs and implementing PDE software for parallel computing. The computational strategy for doing trasient hydraulic tomography on parallel computers is to use the overlapping domain decomposition (DD) ap proach, which is numerically order-optimal and suits inherently for parallel computing. However, many challenges exist in this strategy including, among other things, (1) finding efficient subdomain solvers required in DD, and (2) implementing the softw are capable of achieving high computation efficiency on parallel computers. Moreover, the parallel software developed at Simula must later be incorporated into a cyberinfrastructure for environmental observatories, where basin-scale groundwater surveys c an be carried out in future. The techniques required with respect to software coupling will thus be investigated in collaboration between Univ. Arizona and Simula Research Laboratory. In the process of realizing this long-term goal of building next-genera tion hydraulic survey systems, it is expected that a number of doctor-degree students and post-doc researchers will be educated.