Accumulated volumes of data on water quantity and quality coupled with meteorological data make data-driven analyses of water related problems an effective decision support tool for water resources management. The project is aimed at developing a framework for water resource assessment and management based on machine learning techniques. The framework relies on data which are routinely collected on stream watershed and become available to users almost in real time regime, e.g. water level or precipitation. The study will enhance the methodology and will form a basis for knowledge transfer activities by providing scientifically valid recommendations for improving black box models which can be integrated into early warning systems and used by local authorities for flood management, particularly, in watershed with rapidly changing land use.
The study is conducted on black box models developed following a methodology for short-term hydrological predictions in small watersheds. The methodology utilizes supervised machine learning algorithms and data on streams and watersheds routinely collected by Conservation Authorities.
The project is an integral part of an ongoing research program on developing quantitative techniques for sustainable water resource management at the watershed level.
Year Project Started:
Toronto and Region Conservation Authority
(e.g type 1000 for 1,000)