CEE & ENVE COLLOQUIUM SERIES: Probabilistic Predictions and Uncertainty Quantification with an Analog Ensemble
ENVIRONMENTAL ENGINEERING SPRING 2016 COLLOQUIUM SERIES
Friday, February 12, 2016 • 12:15 PM • Storrs Hall, Room WW16
Probabilistic Predictions and Uncertainty Quantification with an Analog Ensemble
Luca Delle Monache
Scientist, National Center for Atmospheric Research, Boulder, CO
Abstract: The analog of a forecast for a given location and time is defined as the observation that corresponds to a past prediction matching selected features of the current forecast. The best analogs form the analog ensemble (AnEn). The AnEn is a general method to generate probabilistic predictions that has been tested successfully for a range of applications including weather predictions, climate downscaling, renewable energy (wind and solar), air quality (ground-level ozone and particulate matter), and hurricane intensity. The recurring features found across different applications are:
- The ability to use a higher resolution model since only one real-time deterministic forecast is needed to generate an ensemble;
- No need for initial condition or model perturbation strategies to generate an ensemble;
- Reliable uncertainty quantification, i.e., no additional ensemble calibration is required;
- Ability to capture the flow-dependent error characteristics.
- A superior skill in predicting rare event;
Examples of AnEn current applications will be shown, and a discussion on how this technique could be implemented for probabilistic predictions of weather parameters over a two-dimensional gridded domain will follow.
Presenter’s Bio:
Luca Delle Monache is the Deputy Science Director of the National Security Applications Program of the Research Applications Laboratory with the National Center for Atmospheric Research, Boulder, Colorado. He earned a Laurea (M.S.) in Mathematics from the University of Rome, Italy (1997), a M.S. in Meteorology from the San Jose State University, San Jose, California (2002), and a Ph.D. in Atmospheric Sciences from the University of British Columbia, Vancouver, Canada (2006). Before joining NCAR he worked at the Lawrence Livermore National Laboratory. His main interests include probabilistic predictions, uncertainty quantification and ensemble design, urban meteorology, mesoscale numerical weather prediction, ensemble data assimilation, boundary layer meteorology, air pollution, inverse and dispersion modeling, and renewable energy prediction and resource assessment.
To access this seminar’s live broadcast or recording please use the following link: https://mediasite.dl.uconn.edu/Mediasite/Play/107cc29e625744038f830910b0c29a061d
Location of the room: http://maps.uconn.edu/map/locations/148
