University of Saskatchewan

Published purpose

This project aims at developing efficient and easy-to-use models for seasonal forecasting of spring runoff for various watersheds in Alberta, Saskatchewan and Manitoba. The season-ahead forecasts are critical for planning reservoir operations, agricultural water management, flood risk assessments, and prediction of nutrient loadings into large water bodies, such as Saskatchewan River Delta, Lake Winnipeg, and Hudson Bay. An attractive feature of this work is that it will connect mountain headwaters to the Ocean, through Hudson Bay, and will utilize the great potential and prediction skills of rapidly emerging machine learning (ML) techniques. Additionally, correlation and causation analyses will be implemented and their respective outputs will help improve our current understanding of runoff generation mechanisms and controlling factors in prairie landscapes. The expected output of the project will be a set of ML models for seasonal runoff forecasting, along with quantitative estimates of the forecast uncertainty, often neglected in traditional modeling approaches.