93 Model Evaluation Tools (MET) verification statistics visualization

Monday, 24 January 2011
Paul Oldenburg, NCAR, Boulder, CO; and J. Halley Gotway and T. Jensen

Handout (931.3 kB)

The Model Evaluation Tools (MET) was developed by the Developmental Testbed Center (DTC) to provide the numerical prediction community with common software incorporating the latest advances in forecast verification. Recently, the DTC developed a system for storing and visualizing MET verification statistics by integrating database and plotting software. The primary focus of this development was to provide a configurable tool, called METViewer, for use across several different DTC testing and evaluation projects. Evaluation activities that the DTC performed over the past year, for which METViewer was utilized, include the Quasi-Normal Scale Elimination (QNSE) Sensitivity Test, the Hydrometeorology Testbed Collaboration (HMT) and the Hazardous Weather Testbed Collaboration (HWT). It is anticipated that METViewer will be available to the community in the future.

METViewer ingests MET output into a standard SQL database, which facilitates searching and sorting of the verification statistics. Using a user-supplied plot XML specification as input, METViewer gathers and aggregates statistics from the database and generates one or more plots. The three primary types of statistics handled are continuous, categorical and object based. Additional calculations, such as confidence intervals, bootstrap resampling and observation frequencies, are performed during the aggregation and plotting process. The primary focus, to date, has been to produce monthly and seasonal aggregations of statistics plotted over lead time, valid date and precipitation threshold, along with vertical profile plots, for a number of atmospheric variables, including: temperature, moisture and winds. The presentation will describe the tool and provide examples from recent projects.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner