Oops–but maybe not. Or maybe so.
There was a period of time earlier this week when Southern California was under the threat of a powerful earthquake. A heightened swarm of 140+ earthquakes had occurred near the southern end of the San Andreas fault, causing officials to issue a warning that there was between a 1/100 and a 1/3000 possibility of an earthquake within the next few days.
Happily, that didn’t happen. Unhappily, pressure on a fault line doesn’t go away by itself, so at some point, the forecast earthquake WILL happen. So what’s a forecaster to do?
The dangers of making forecasts
There are basically two dangers when one attempts to make a forecast of a natural disaster:
1) The forecast will come true. People will be killed and injured, and property will be destroyed and damaged. However, forecasters will pat themselves on the backs because, from their perspective, the situation could have been much worse–and this is probably true.
2) The forecast will not come true. People will not be killed, property will not be destroyed, but the forecasters will take a credibility hit of damaging magnitude, and the even could cause the apathetic public to ignore and disregard future forecasts. Then, when the forecast ultimately does come true, people will be killed and injured, and property will be destroyed and damaged, at a rate higher than if earlier forecasts had been accurate.
So in the previous example, the operative question is this: Because the earthquake did not occur, how many additional people will ignore the next warning? How many people will be injured because of that? How much property will suffer additional damage because of that?
Natural disasters are not all the same.
It’s important for you and your served pubic to understand this. Forecasts for hurricanes, floods, tornadoes, and earthquakes couldn’t be more different. But if you don’t couch the forecasts in terms of relative certainty, then your public could quickly lose confidence in you. So let’s go through them one by one.
As this is being written, Hurricane Matthew has passed through Haiti and is approaching Bermuda. Although the forecast models are sometimes humorously inaccurate with regard to long-term path, they tend to be absolutely accurate in the near term–one or two days.
So, if the forecast says you’re going to get wet, then you’re going to get wet. If it says your trees will blow over, then they will. If it says that 20 feet of surge will inundate your 15-foot tall house, then that’s what’s going to happen. You can pretty much take that to the bank–or in this case, your insurance company.
The issue with floods is that the cause often happens far away from the effect. In other words, torrential rain in South Dakota can cause debilitating floods in Louisiana–as has happened before. When a wall of water is moving downstream, there’s little to no uncertainty at issue.
Again, if the forecast says that tomorrow, the water will be 10 feet deep where you’re standing, then that’s what’s going to happen. Tomorrow, you’d better not be standing there.
Here’s where it starts getting more uncertain. The Weather Channel has done some great work in tornado forecasting, which has resulted in the TORCON system. Essentially, if the forecast is that the TORCON for the day is 3, that means there is a 30 percent chance that a tornado will form within 50 miles of your location.
Development of this system is quite a remarkable achievement, and, as a resident of tornado alley, I can testify that that it’s pretty darn accurate. Whenever a TORCON of 5 is issued, I’ve found that about half of the time, there actually IS a tornado that forms within 50 miles of my house.
However, as remarkable as the system is, it can be argued that it’s nearly irrelevant. I’ve been under a TORCON warning many times in the past few years, and no tornado has ever touched down anywhere near me (not that I’m complaining). But the point is that if a forecast can’t tell me for sure that a tornado will hit my house, I’m going to start giving it less credibility and importance. That puts tornado forecasting in a tough spot. A 50-mile radius is big, and a tornado is small–often just a few hundred yards wide, and very short-lived–so as valuable as it is to know that one may form, it really doesn’t help protect life and property with any surety.
Earthquakes present the toughest scenario of all to the forecaster. Will an earthquake happen? Check. Will it happen right here? Check. Will it happen tomorrow? Not-so-check. Will it happen in the next 100 years? Maybe check.
When doing earthquake forecasts, we’re talking about formations of rock pressing against other formations of rock 10+ miles below the surface. It’s no wonder that the forecast accuracy is so small.
But for the same reasons that the tornado forecasts aren’t useful, the earthquake forecasts aren’t useful either. If I need to shelter my family, I need to know when to do that. If I need to shelter my family, I need to know where the affected area is. We can’t give that information to our served public, because we don’t know it.
Maybe someday we will, but we don’t now.
Include Accuracy and Probability in your Forecasts.
In the bottom line, natural disaster forecasters have a tough job. Part of it is fairly easy and fairly straightforward–particularly with respect to hurricanes and floods. Part of it is much more complicated and imprecise–tornadoes and earthquakes.
Yet our served public, if not educated to these differences, will assume that all forecasts are accurate until they’re not. Once they’re not, then they will tune out warnings that could save their lives.
So it’s important to note IN ADVANCE when a forecast is going to be accurate and when it’s likely to not be accurate. This caveat in your presentation could make all the difference in the world.