The False Alarm Rate is the percentage of non-precipitation days that were incorrectly forecast as wet: $F / (F + CN)$, where F = false alarms and CN = correct negatives. It is also known as the “probability of false detection,” and a lower value is always better.
This is a common point of confusion in the industry. The Rate looks at all the days it didn’t rain and asks, “How often did you wrongly say it would?” The Ratio looks at all the times you said it would rain and asks, “How often were you wrong?” Professionals search for “False Alarm Rate vs. Ratio” to ensure they are accurately communicating model bias to stakeholders.
If a concrete contractor cancels a pour because of a rain forecast that turns out to be a “False Alarm,” they lose thousands of dollars in labor and materials. They search for providers with a low False Alarm Rate to minimize “unnecessary downtime.”
While the term is primarily used for precipitation, it can be applied to any binary event, such as “Will the temperature drop below freezing?” A high False Alarm Rate for frost can lead to “Alert Fatigue,” where farmers stop protecting their crops because the previous five warnings were “False Alarms.”