The Model Evaluation Tools: How to use its statistical output to understand model performance

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 6 February 2014: 3:45 PM
Room C202 (The Georgia World Congress Center )
Tara L. Jensen, NCAR/RAL, Boulder, CO; and T. L. Fowler, J. H. Gotway, B. G. Brown, and R. Bullock

The Developmental Testbed Center's (DTC) Model Evaluation Tools (MET) has become a cornerstone tool in the DTC testing and evaluation process. It is also used world-wide by over 2,000 registered users. In-depth examples of new capabilities within MET will be presented and will cover the Series Analysis tool, pre-processing of high temporal resolution observational datasets, and the calculation of spread-skill within the ensemble tool. This talk will also go beyond the traditional introduction of capabilities available in MET and explore how model developers and evaluators can use MET's less-utilized statistics and metrics to understand their model performance (i.e. odds ratio, probability integral transform, ignorance score and others).