A skill score measures the accuracy of a forecast with reference to the accuracy of a standard forecast.
The standard forecast is usually but not always a forecast that is available to a forecaster, and which
does not require any effort or knowledge on his part to prepare.
For example, a forecast which consists of the climatological average temperature for a particular station
for each day can be obtained from climatological tables and doesn't require any knowledge of the current
weather situation to prepare.
A skill score is a comparison of the score obtained by a forecast with the score obtained by the standard
forecast using the same set of verification data.
The format of skill scores is usually the same:
The score used may be any of the common accuracy scores used in verification.
For continuous variables, skill scores are usually based on either the mean absolute error or the
mean squared error.
For probability forecasts, skill scores are usually based on the Brier score.
The skill score in this format measures the percent improvement of the forecast compared to the standard.
Interpretation:
The skill score has a range of -∞ to +1.
A positive value of the skill score means the forecast is an improvement over the standard forecast.
A negative value means that the forecast has lower accuracy than the standard forecast,
and one might as well use the standard forecast instead.
Loading Questions
...
You are told that your skill score for
temperature forecasts last month was +0.25 (25%).
The standard forecast used for comparison was in error by
2 degrees on average.
This means:
A skill score is the percentage improvement with respect
to the standard, not to the difference.
Your skill is positive.
Therefore you must have a smaller error than climatology.
Yes. Your skill is 25 percent better than climatology.
25% of 2 degrees is 0.5 degrees.
