As a result of the considerable insured damages resulting from flood events, the development of flood prediction models to identify flood-prone areas and predict the timing and severity of flooding events is of great importance to better prepare for and mitigate adverse effects of flooding. The application of hydrologic and hydraulic models can be limited due to the extensive degree of input data and their required temporal and spatial resolutions. Considering this, the use of data-driven, statistical machine learning models to better understand the nature of flooding and the performance of flood prevention measures is a promising alternative.
This research applies and compares the performance of various statistical and machine learning models for flood susceptibility mapping, to better predict flooding events, and to evaluate the performance of lot-level flood mitigation strategies. This research can help identify key factors related to local flooding in urban environments, the timing and severity of flooding events, and lead to improved knowledge of flood mitigation measures to better protect homes from damage due to flooding.
Elkurdy, M., Bonakdari, H., Binns, A.D., Gharabaghi, B., McBean, E.A. Early detection of riverine flooding events using the Group Method of Data Handling for the Bow River, Alberta, Canada. Submitted to International Journal of River Basin Management.
Conference presentations and abstracts:
Elkurdy, M., Binns, A.D., Gharabaghi, B. Improved streamflow forecasting using variational mode decomposition and extreme gradient boosting. Submitted to AGU Fall Meeting 2020, 7-11 December, San Francisco, USA, 2020.
Kaur, B., Szentimrey, Z., Binns, A.D., McBean, E., Gharabaghi, B. Urban flood susceptibility mapping using supervised regression and machine learning models in Toronto, Canada. Submitted to AGU Fall Meeting 2020, 7-11 December, San Francisco, USA, 2020.
- Results can be used to identify flood susceptible areas and help to provide rapid emergency response services
- Results can guide in implementing targeted flood prevention strategies at the lot-level
- Results can help to understand how future changes in conditions may affect the performance of lot-level flood mitigation strategies (e.g., backwater valves, foundation drainage systems).