Ed Sudicky

University of Waterloo 
Project Title: Developing real-time hyper-resolution simulation capability for the HydroGeoSphere (HGS)
Industry Partners: AquantyIBM Canada Ltd.
Project Partner: David Lapen
Project Title: Real-time cloud-based hydrologic risk assessment platform development for watershed scale applications
Platform: Cloud Analytics

Digital Media Water

Developing real-time hyper-resolution simulation capability for the HydroGeoSphere (HGS)

Climate change will greatly impact the availability and quality of Earth’s water resources over the next century. The expected increase in mean temperature will have a severe impact on the water cycle, not only through changing precipitation patterns and amounts, but also through an increase in the severity and frequency of extreme events. Already, changing rainfall patterns and shifting temperatures are increasing the complexity of water management in the Grand River watershed and affecting the programs and operations of the Grand River Conservation Authority (GRCA). Rigorous science-based forecasts to address how the surface and subsurface (groundwater) resources might be impacted by climate change will therefore necessarily demand the use of a computational platform that fully integrates the climate system with the surface/subsurface hydrological system in three dimensions. By establishing proactive science-based management policies now, such as water use quotas, limits on fertilizer and pesticide use, water treatment guidelines, flood control practices, etc., the future sustainability of the water resources can be protected, and perhaps even enhanced. The high-resolution regional climate simulations being performed by Prof. W.R. Peltier’s group will provide the data to drive our 3D integrated surface/subsurface hydrological model. We are also co-ordinated with the smart data collection activities being undertaken in the Southern Ontario Water Consortium (SOWC), of which IBM is a major partner.

Real-time cloud-based hydrologic risk assessment platform development for watershed scale applications

Real-time hydrologic modelling with physics based tools such as HydroGeoSphere (HGS) requires a HPC platform that can facilitate multiple concurrent simulations that have inter-simulation data sharing capability, as well as the ability to feed real-time input data to the simulations while also extracting and visualizing real-time model output data. Typically, a watershed scale HGS model requires 8 to 12 cores on a shared memory platform for a single simulation to run efficiently, and for real-time simulations it is expected that at least 3 to 5 simulations will need to run concurrently. Furthermore, each simulation can easily generate 250 Gb of model output data, which leads to large data storage and data manipulation requirements when the output from numerous simulations is combined into a single database. While the technical challenges are significant, a real-time HGS simulation platform will be the first physics based modelling tool available for use by water resource managers and water related risk specialists from many fields, including agriculture, insurance, transportation and infrastructure management. The potential market for this type of modelling platform is global in scale.