April 19th, 2023 12pm ET: CWRA SYP Ottawa Webinar: “Machine Learning applications in Hydrology and Geomorphology” with Hannah Burdett and Cody Kupferschmidt

#1 Presentation Title: Forecasting River Channel Migration – Combining Existing Approaches with Machine Learning

Speaker: Cody Kupferschmidt

Abstract:

The geomorphic processes that cause river channel migration can be highly variable and result in significant hazards. Being able to accurately and confidently predict future changes in river planform geometry plays an important role in helping to mitigate risks. In this presentation, we provide a review of existing techniques for predicting river channel migration including aerial imagery/survey-based methods, empirical/probabilistic methods, and physically-based methods. We identify potential use cases for machine learning to help improve these processes, and present preliminary findings from our research conducted on the topic.

Speaker’s Bio:

Cody Kupferschmidt is a Ph.D. Candidate at the University of Guelph, where he is currently working with Dr. Andrew Binns in the School of Engineering. Cody’s current research is focused on the applications of machine learning to water resources engineering, with an aim to improve predictions for river channel migration. Cody is also a registered professional engineer (Ontario), and prior to returning to school to pursue his Ph.D. worked in the consulting engineering and renewable energy sectors.

#2 Presentation title: Deriving Snow Ablation Upscaling Relationships via Machine LearningSpeaker: Hannah Burdett

Abstract:

This presentation will provide an overview of the development of an upscaling methodology that examines the temporal and spatial variability of sublimation and snowmelt fluxes in a drainage basin in the Canadian Rockies through machine learning methods. Although hydrologic processes are often measured and understood at the point scale, there is significant spatial variability across larger landscapes. This study involves estimating spatially averaged results from a fine-resolution discretized hydrologic model without explicit knowledge of detailed local response and with and without low-order statistics of state (e.g., the standard deviation of snow water equivalent). A series of experiments progressively increasing in complexity are used to test and validate that the upscaling methodology can successfully represent the impact of heterogeneity within the system.

Speaker’s Bio:

Hannah is a Ph.D. candidate at the University of Waterloo under the supervision of James Craig, where she applies machine learning methods to assist with upscaling in hydrologic modelling. She previously earned a Master’s in Spatial Analysis from Toronto Metropolitan University, where she used machine learning applications to predict crop yield. In addition to her current research, Hannah has developed Magpie, an open-source Python and R workflow created in Google Colab to assist with geospatial data preparation and model construction for Raven, a flexible hydrological modelling framework.