Speaker : Omar Wani, PhD Researcher, ETH Zurich and Eawag, Switzerland
Title of Talk: Uncertainty Analysis for Environmental Models - An Introduction.
Date: 20 March, 2017 (Monday)
Time: 04:00 p.m. - 05:00 p.m.
Venue: First Floor Seminar hall, CE.

______________________________________________________________________________


Speaker Bio: 
************
Omar Wani has a BTech in Civil Engineering, with emphasis on Hydraulic
Engineering, from Indian Institute of Technology, Roorkee (2012). He received the EU
Erasmus Mundus scholarship and went on to pursue his Masters in Water Sc. & Eng. from
UNESCO-IHE, the Netherlands and Hydro Sc. & Eng. from Technical University Dresden,
Germany (2014). He is currently an EU Marie Curie Fellow, working on his doctoral studies,
at the Institute of Environmental Engineering, ETH Zurich and Eawag, Switzerland. Being part
of the QUICS (Quantifying Uncertainty in Integrated Catchment Studies) consortium, his
research focuses on proper quantification and reduction of uncertainty for environmental model
forecasts. Prior to his doctoral research, he worked on the development of uncertainty analysis
tools for Flood Early Warning System (FEWS) software package at Deltares, the Netherlands.


Brief abstract: 
**************
Accurately predicting changes in water quantity and water quality is important
for managing hydrologic catchments, be they urban or rural. Due to the complexity of
hydrologic processes, with process variability at several scales, deterministic forecasts for
natural or engineered catchments are seldom accurate. Uncertainty quantification helps to
better assess the accuracy of model predictions and helps improve the predictive capabilities
of our models. This talk will discuss the use of Machine Learning (ML) and Gaussian processes
as uncertainty descriptors in hydrologic modelling.

Special ML techniques can be used in the post processing phase of the forecast for uncertainty
estimation. Also, Gaussian processes can be used to explicitly add stochasticity to models.
After such a description of uncertainty, parameter estimation can be carried out within a
Bayesian framework. Application of these uncertainty estimation techniques on various
catchments will be discussed during the presentation.