The big difference between earthquake prediction and forecasting

I previously posted about how it’s unethical to endorse dowsing if you are a geologist bound by a professional code that includes using the best scientific procedures and evidence. Condoning a process which is scientifically questionable or invalid is a breach of this code.

A similar argument can be made for earthquake prediction. There have been several instances where scientists (and many more non-scientists) have predicted through various means when and where an earthquake will occur. Currently, there is a storm of criticism leveled at author (not scientist), Simon Winchester after he wrote this article strongly suggesting without evidence that the Pacific coast area is next in line for a big quake due to the strain at “a barely tolerable level” (whatever that means).

These next two blog entries will explore natural disaster prediction. First, it’s important to distinguish between prophesizing, predicting and forecasting.

I wrote about psychics’ failure to predict natural disasters. I’ll be very clear — psychic prophesizing just doesn’t work.

Claims to predict earthquakes have been based on animal behavior, patterns of foreshocks, radon gas levels, ionospheric anomalies, and psychic visions. They really run the range from ridiculous to plausible; from the metaphysical to the scientific. What is the public to make of this? How do you know whom to believe?

Prediction. The primary geologic institution of the U.S. and one of the world’s most respected and consulted agency is the U.S. Geological Survey. They employ the world’s foremost experts on seismology and other geologic specialties. The USGS’s stand on prediction of earthquakes is that it is not currently possible. It is not for lack of trying. They have tested many theories but none currently developed are reliable. My next post will give example of scientists who attempted to predict.

Yet, earthquakes can be forecast. What’s the difference? I found this good explanation on an earthquake forecasting site:

“While the dictionary defines forecasting to be a synonym for prediction, we consider forecasting to be a specification of the odds, or probability, of an earthquake occurring at a given location, during a given time window, within a given magnitude range.  By contrast, we consider a prediction to be the specification that an earthquake either will, or will not, occur at a given location, during a given time window, within a given magnitude range.  A forecast is therefore a statement of probability, whereas a prediction is a binary statement.  An individual forecast can never be validated by a single observation, but a forecast method can be validated by many observations.  By contrast, an individual prediction can be validated by a single observation.”

In forecasting, we get probabilities, not “this place, this day, this time”.  As scientists will tell you, that level of specificity is too much to ask considering the variables of nature. So, forecasting earthquakes is much more reasonable considering the data that is available. Think of it as a weather forecast or a hurricane forecast – you know what’s coming based on the data but you do not know quite enough to pinpoint exactly how things will progress. You can get a reasonable idea and, more importantly, you can prepare for it. As more data is fed in, there is the potential that you can get a more accurate forecast.

Models are used to forecast. Models are constructed simulations of how a natural process works. By specifying parameters, such as equations, boundaries and other data into a mathematical model, it can be used to predict what might happen in the future if you change one variable (add more water, higher temperature, etc.) or even to determine what happened in the past (that is, you can run a model backwards in time as well as forwards). Very handy things, models. However, a model is only as good as the input. In other words: “Garbage in — Garbage out”. Good scientists tweak and validate their models. They can test them on real world situations where they know the outcome. Do the model’s predictions match what really happened? Yes? Then the fit is good. No? Something is not quite right. Back to tweaking.

There have been several times in history where respected scientists or just media savvy individuals have made earthquake predictions. In these cases, the individuals had gone outside of established scientific knowledge, out on a limb you might say, and stated their claim that an earthquake will occur in a precise place at a specific time. Typically, they feel that their prior predictions have been validated through their own testing. But models either did not exist or were not good enough yet.

To gain support for one’s theory, the established method is to test and then publish in a scientific journal. Then, your peers can view your work and critique it for errors, bias or things you missed. In a follow-up to this discussion, I resurrected a piece I wrote on the ethics of natural disaster prediction that expands upon the credibility and responsibility of scientists in such situations. And, I provide real world examples of predictions gone wrong. They went wrong for many and various reasons. It’s a frustrating topic. It also won’t be going away soon.