Webinar – Groundwater and Surface-Water Forecast Evaluation in Two Contrasting Hydrostratigraphic Settings within Southern Ontario
AUTHORS: S.K. Frey, G. Stonebridge, H. Zhang, O. Khader, A.R. Erler, N. Shamilisham, David Hah & E.A. Sudicky
We are happy to see the 2022 Southern Ontario Groundwater Geoscience Forum content is now available online (see below). On February 15th Aquanty’s Steve Frey presented our latest work associated with the project for a fully integrated groundwater–surface-water model for southern Ontario.
In this presentation we highlight our efforts to couple two watershed scale HydroGeoSphere models with the latest weather forecasts. The results allow us to forecast future groundwater levels and surface water flow rates over the short to medium term (i.e daily to monthly).
Click here to view the full list of talks presented at the 2022 Southern Ontario Groundwater Geoscience Forum.
ABSTRACT:
Over the course of the 2014–2019 Southern Ontario Groundwater Project, a regional-scale HydroGeoSphere (HGS) fully integrated groundwater–surface water model was developed and tested. From the regional model, a derivative set of watershed-scale HGS models were constructed with much higher levels of spatial resolution. The watershed-scale models have been incorporated into a hydrologic forecasting system that provides coverage across southern Ontario. In 2019, a project was initiated to evaluate the ability of the forecasting system to predict future groundwater levels in the Long Point and Quinte regions, which are both prone to groundwater stress during drought conditions. As the hydrostratigraphic settings of Long Point and Quinte are considerably different, comparison of simulation results between the two regions provides insight on structural factors that influence forecast skill. The forecast evaluation spans the summer and fall of 2021 and includes 7-day and 32-day forwardlooking forecasts that were generated at respective daily and weekly frequencies. The skill metrics are calculated for both surface water flow rates and groundwater levels, based on comparison between forecasts and subsequent observed conditions at Water Survey of Canada hydrometric stations and Provincial Groundwater Monitoring Network real-time monitoring wells.
Results from the skill analysis indicate that groundwater levels tend to be much more predictable than surface water flow rates. Based on Kling-Gupta efficiency (KGE) scores, root mean square error (RMSE), and bias, surface water forecasts tend to exhibit a notable degradation of skill between 3 and 5 days, with seasonality playing a role. In contrast, groundwater forecast evaluation based on linear correlation between simulated and observed groundwater levels, RMSE, and bias, indicates that skill extends beyond 20 days at most PGMN well locations utilized for the analysis and beyond 30 days at others, with less influence from seasonality. Groundwater forecast skill also tends to extend longer for the Long Point region compared to Quinte, although the lower number of real-time PGMN wells in Long Point may be influencing this interpretation. The work presented here demonstrates that groundwater forecasting in southern Ontario is a viable tool to help anticipate and manage groundwater resources.