Streamflow monitoring is critical for effective water resource management, environmental monitoring, flood mitigation, and ecosystem services. However, there is an increasing decline in streamflow measurements in Canada. In 2023, the total hydrometric stations operated by Water Survey of Canada were 20% lesser than in 1980s, owing to funding constraints, aging infrastructure and competing interests. At the same time, advances in new technologies, such as remote sensing and drones, and focus on data analytics through predictive modelling and machine learning are believed to be able to compensate for lack of streamflow measurements. In this study, we implement hydrological modelling, machine learning and statistical approaches to generate a proxy streamflow data for a catchment, where we have long-term observed data available. Then we perform several analyses, including flood frequency analysis, flow duration curve, trends and indicators of hydrological alterations to show that field observations can be complemented, but not substituted by alternative approaches.
