Water flow in prairie watersheds is increasingly unpredictable — but AI could help

The Prairies experience unpredictable climate swings, creating challenges in flood preparedness and water quality management due to their unique landscape.
The Prairie Pothole Region features numerous shallow wetlands that store water before spilling into streams, complicating streamflow prediction.
Limited streamflow monitoring across the Canadian Prairies means many watersheds lack gauges necessary for reliable water level and flow rate measurements.
Variability in wetland water storage significantly affects streamflow in the Prairie Pothole Region, making predictions complex and challenging.
A new study combines Prairie Pothole physics with artificial intelligence to estimate streamflow and wetland water storage in unmeasured areas.
The AI model integrates the fill–spill–connection physics of the wetlands, predicting regional patterns that work in areas without streamflow records.
Successful tests across 98 watersheds show the model’s reliability in forecasting streamflow dynamics and capturing wetland storage fluctuations.
The innovation provides practical benefits such as enhanced flood preparedness, better understanding of regional storage variability, and linking hydrological knowledge with modern data-driven tools.