Three steps to simulation
With the underlying datasets assembled into a single integrated model, Canada1Water is now in its spin-up phase. Hydrogeologist Omar Khader describes what’s involved in getting the models ready to run continental-scale simulations.
THE SPIN-UP PROCESS
Approx. 6 weeks Approx. 26 weeks
Over the coming months, Khader and his colleagues will put the Canada1Water model through a three-step process of sensitivity analysis, calibration and validation.
“By the end of spin-up, we will have achieved dynamic equilibrium,” Khader explains. “That means model outputs like stream flow reflect seasonal variations with the values remaining consistent from year to year, and lake levels, groundwater levels and other features are as close as possible to the observed record.”
Sensitivity analysis: What affects the model most?
The first spin-up step involves isolating individual model parameters and testing their impact on overall model performance. This determines which parameters will need the most attention during calibration.
Given Canada1Water’s continental scale and complexity, some simplification is required to produce a computationally feasible simulation. There are dozens of parameters to consider: surface water friction and stream geometry, groundwater conductivity and storage, plant root depth, soil evaporation depth, porewater pressure limits for evaporation and transpiration — and many more.
“We can’t resolve all the stream orders in the model, for example,” Khader says, “but we can include the main streams that affect the big picture. Similarly, we can’t resolve every single geological feature, so we’re simplifying subsurface layers and assigning properties that suit the overall level of detail across the study area.”
“Canada1Water extends from the Great Lakes to the peatlands of Hudson Bay to the Great Plains and Rocky Mountains. In every case, what we need to calibrate is going to be a little different.”
If any of these simplifications ends up affecting model performance, corrections will be made at the next stage — calibration.
Calibration: Fine-tuning the model
Dr. Omar Khader is a Hydrogeologist at Aquanty responsible for Canada1Water model assembly and calibration.
Calibration involves testing the various parameters and making sure they align with real-world measurements. Khader and his colleagues will set minimum and maximum values for each parameter, then assign values from within those ranges and assess how the model performs.
While in principle this sounds straightforward, it is complicated by the fact that different parameters could affect model outputs in different geographic regions, presenting challenges for the calibration process.
“Canada1Water extends from the Great Lakes to the peatlands of Hudson Bay to the Great Plains and Rocky Mountains,” Khader says. “In every case, what we need to calibrate is going to be a little different.”
For example, in mountainous areas where water flows quickly, simulating peak flow values and timing is harder than in other areas, requiring specific calibration approaches. The Prairies, on the other hand, present almost an opposite issue: shallow wetlands known as ‘potholes’ store surface water in ways that affect groundwater recharge and seepage, and the model has to be fine-tuned to account for these effects. Peatlands are another case altogether, affected by the hydraulic properties of topsoil and by evapotranspiration in ways that make it hard to model the true extent of water in the system.
Resolving each of these challenges will bring the complete model closer to a state of readiness to run accurate predictive simulations.
Validation: The final performance check
The final spin-up step involves running the calibrated model in a different time window than was used for calibration. If the two runs have similar accuracy when compared to historical observations, the team can take confidence in the calibrated model. If there are discrepancies, they’ll go back into the calibration process and make further adjustments.
“It’s extremely iterative and, on this scale, enormously complex,” says Khader. “But in the end, we’ll know we have the world’s first comprehensive physical model of a continental groundwater–surface water system that can be relied on to make long-term planning decisions.”