5.1 Verif: A Python Package for Forecast Verification

Tuesday, 24 January 2017: 4:00 PM
Conference Center: Chelan 5 (Washington State Convention Center )
Thomas Nipen, Norwegian Meteorological Institute, Oslo, Norway; and D. M. Siuta and R. B. Stull

Handout (2.2 MB)

Verif is a tool to compute and visualize verification scores for point forecasts. It can be used to document the quality of one forecast system but also to compare output from multiple competing weather models or post-processing schemes. It is an open-source command-line tool written in python and can be downloaded at https://github.com/WFRT/verif .

The package contains more than 80 metrics and diagrams and includes deterministic scores such as mean absolute error, threshold-based metrics such as the equitable threat score, probabilistic scores such as the Brier score, and composite diagrams such as the Taylor diagram. Scores can be graphed as a function of date and forecast lead time, but can also be plotted on a map to show spatial variability of the scores.

The program requires the user to put observations and forecasts along with their initialization time, lead time, and geographical coordinates into a simple text format. Forecasts can be in deterministic, ensemble, or probabilistic form.

The tool allows quick on-the-fly investigation of forecast performance, but can also generate figures automatically in an operational setting. A range of options exists to control image dimensions, colors, and font sizes.

- Indicates paper has been withdrawn from meeting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