Probabilistic Skill Scores

Probabilistic Skill Scores

Some of the previously mentioned verification statistics for probabilistic forecasts can be used to compute skill scores that are regularly used. All of these skill scores use the same general equation format of comparing their respective score results to a base or reference forecast, which can be climatology, a “random” forecast, or any other comparison that could provide useful information.

Brier Skill Score

The Brier Skill Score (BSS) measures the relative skill of the forecast compared to a reference forecast. It’s equation is simple and follows the general skill score format:

Compared to the BS equation, the range of values for BSS differs: the range is from negative infinity to 1, with 1 showing a perfect skill and 0 showing no discernable skill compared to a reference forecast. See how to use this skill score in METplus!

Ranked Probability Skill Score

Similar to the BSS, the Ranked Probability Skill Score (RPSS) follows the general skill score equation formulation, allowing a comparison between a chosen reference forecast and the forecast of interest:

RPSS measures the improvement/degradation of the ranked probability forecasts compared to the skill of a reference forecast. As with BSS, RPSS ranges from negative infinity to 1, with a perfect score of 1, and a score of zero indicating no improvement of forecast performance relative to the reference forecast. See how to use this skill score in METplus!