Machine Learned Emulators: Discovering Teleconnections and Quantifying Uncertainty in Coupled Socio-Hydrological Systems
Day 3 Session 2 (1:30PM)
Presenters:
Asim Zia, University of Vermont
Patrick Clemins, University of Vermont
Muhammad Adil, University of Vermont
Actionable water intelligence requires quantification of uncertainty in hydrological prediction as well as clear identification of drivers of change both internal and external to the hydrological system of interest. Traditional approaches to quantify uncertainty and explore teleconnections in process-based models of coupled socio-hydrological systems range from global sensitivity analysis of model parameters to Monte Carlo simulation experiments, decomposition analyses, and error propagation methods.
This workshop will provide theoretical, computational, and water management policy examples of machine learned emulator models for discovering teleconnections and quantifying uncertainty attributable to internal versus external drivers of change.
Learning Outcomes:
- Understand the need for and methods used to quantify uncertainty for actionable water intelligence derived from hydrological models
- Train and test machine learned emulator models to quantify uncertainty in coupled models
- Post-process machine learned emulator models to discover teleconnections and distinguish internal versus external drivers of change in coupled socio-hydrological systems
Prerequisites:
- Accounts Required:
- GitHub