The Brier Skill Score measures how much better a provider’s probability forecast is relative to a reference forecast (usually “Persistence” or “Climatology”). A higher BSS indicates more skill; a BSS of 0 means no improvement over the baseline, and a negative BSS means the forecast was worse than simply guessing the average.
Many apps claim 90% accuracy, but if the climatological “chance of rain” is only 10%, a “0% rain” forecast will be “accurate” 90% of the time without any skill. The BSS strips this away. Industry professionals search for “BSS vs. Climatology” to see if a weather provider is actually adding value or just repeating the historical average.
“Persistence” is the forecast that says “tomorrow will be exactly like today.” In stable summer patterns, persistence is hard to beat. A positive BSS against persistence means the forecaster correctly identified a change in the weather pattern (e.g., a cold front arrival). Pros search for “BSS persistence benchmarks” during transitional seasons like spring and fall to see which models handle “pattern breaks” best.
A negative BSS is a red flag. it means the forecaster’s “probabilistic” logic was so flawed that a user would have been better off just looking at a historical climate chart. This often happens in models that “over-forecast” rain in the tropics, where the BSS can drop significantly if the model ignores the “Uncertainty” of local afternoon showers.