So I’m not saying I know the proper way to calculate it, but I’m not convinced yours is that way.
The way I did it is not great. It’s simple, and verifiable, but I would not call it infallible in the least.
If we do this with rain events, we don’t just use one rain gauge. We use a wide array of rain data across a geographic area, and we amortize it, and we draw maps based on a kind of weighted geographical averaging. They’re called “isohyetal maps,” and they would show that the rare events which hit Florida drop more rain than the rare events that hit Georgia, for instance. That’s a very sound mathematical procedure because we have comps. Irregularities at individual data points come out in the wash.
What I did was basically take rain events from one weather station in isolation. It’s not a great approach, but it’s the beginning of an approach.
If we really wanted to do this right, in my estimation, we’d do it culturally, and we’d do it on a global scale. So arabic cultures might for instance have a higher P(R) than nordic cultures. Latino cultures might have a different number, etc. And then we could try and extrapolate what the natural inclinations the USA might have for P(R) based on our ethnic mix.
We also might run the same analysis for different economic structures or such, and see which one is more predictive.
That sort of analysis is way outside the scope of the article though. In the end, quarreling about the P(R) number misses the point of the article, which is merely to say that there is a P(R) number, and that die gets cast once a year. We live a long time. I’m totally open to someone else developing their own P(R) number and trying the math. You could develop your own, and it might be a better one that I cooked up with my simplified analysis. When you plug it in, you’ll very likely see that the chance of getting swept into one of these things is more than you think. Try it out.