Utilizing Extensive Monitoring Data to Evaluate Green Stormwater Infrastructure Performance

Abstract Submission: Performance assessment of green stormwater infrastructure (GSI) is essential to ensuring sustainable urban stormwater management. However, this task is challenging due to the complexity of the systems, as well as financial and workforce constraints. GSI performance predictions are often based on municipal design regulations such as ponding duration or overflow. Predictions of performance is typically done through physics derived hydrological modelling, requiring either simplifications of the highly dynamic nature hydrology of these systems or complex data inputs and high computational costs. Recent advances in data-driven approaches, machine learning (ML) have created a new opportunity to improve prediction of GSI performance. This study applies different ML models to predict ponding duration and maximum ponding depth of storm water in GSI systems. A database of stormwater observations for different type of GSIs was used with weather data, to develop predictive models for GSI performance. Individual storm events were identified from the observed dataset, and features such as precipitation depth, storm duration, average intensity and maximum intensity were extracted for each storm. Individual ML models were trained and tested using those features and then other models were developed with combined dataset from GSI sites along with GSI drainage area, footprint, type. Storm event data coupled with GSI site description was able to predict ponding times. These findings are an important stride towards sustainable and efficient management of GSI at the city-scale.

Learning Objectives/Expected Outcome (Optional) :