Automatic model calibration for improving numerical weather forecasting

Qingyun Duan is currently a professor of hydrology and water resources at the Institute of Land Surface System Science and Sustainable Development of Beijing Normal University (BNU) in China.

Special Seminar on Monday, February 5, 2018 at 4 pm in Harshbarger 206 - Refreshments at 3:45

Automatic model calibration refers to the process in which the parameters of a dynamical model are tuned using mathematical optimization methods to minimize the aggregated difference between model predictions and corresponding observations. This approach is not widely practiced so far in numerical weather predictions because of difficulties related to model complexities such as high-dimensionalities of model parameters and model outputs, and the extraordinary demand of computational resources. This talk introduces a platform called Uncertainty Quantification Python Laboratory (UQ-PyL) to perform automatic calibration of a numerical weather prediction model as the WRF model. The key functions of UQ-PyL include design of experiment (DoE), uncertainty analysis, global sensitivity analysis, surrogate modeling, and multi-objective optimization. Those functions will be demonstrated using the WRF model with a case study involving 5-day weather forecasting in the Greater Beijing region. We found that automatic model calibration can improve predictive skill of the WRF model significantly according to numerous skill metrics.

Bio:

Qingyun Duan is currently a professor of hydrology and water resources at the Institute of Land Surface System Science and Sustainable Development of Beijing Normal University (BNU) in China. Prior to his current position, he worked at U.S. NOAA Hydrology Laboratory from 1991 to 2003 and U.S. Department of Energy Lawrence Livermore National Laboratory from 2004 to 2009. Dr. Duan obtained both his MS and Ph.D. degrees from the Department of Hydrology and Water Resources of the University of Arizona. His research interests include: hydrology and water resources, hydrological model development and calibration, hydrometeorological ensemble forecasting, and uncertainty quantification for large complex system models. Dr. Duan has been active in many international scientific activities, including serving as the leader of the Model Parameter Estimation Experiment (MOPEX) and a member of the scientific steering committees of the Global Energy and Water Exchange (GEWEX) Project and the Hydrological Ensemble Prediction Experiment (HEPEX). He was or is serving as an editor or editorial board member for numerous scientific journals, including Bulleting of American Meteorological Society and Water Resources Research. Dr. Duan is a recipient of Chinese Government One-Thousand Talents Program award, a Fellow of American Geophysical Union and American Meteorological Society.


File Upload