Sherlock Holmes, the great detective, was also an expert statistician. At least, that’s what recent lecturer and statistician Dr. Douglas W. Nychka claims.
On October 24, Nychka visited BYU to participate in a seminar held by the Department of Statistics. During the seminar, he shared his thoughts on how statistics could be a potential answer for more accurate weather predictions. Nychka claimed that weather prediction could be a lot simpler if we use methods similar to the deductive reasoning Sherlock Holmes used to solve his mysteries.
“How did Holmes do it?” asked Nychka. “It’s a simple combination of prior knowledge combined with observation. Much like Holmes would notice dirt on someone’s collar, we can use satellite images, weather balloons, rawinsondes, and more to measure the entire vertical profile of the atmosphere.”
To apply this method of weather prediction, we take what we know of the past, our prior data, or a sample of the state of atmosphere over time. This can then yield more specific and more accurate weather predictions of the future.
However, this process isn’t necessarily easy or even possible for the average human brain to accomplish.
“Take this method that we just did to predict the weather for one tiny dot on the globe,” said Nychka, “and multiply it by the quarter of a million grid points we have set across the world, and you’ll understand why we need a super computer.”
At any particular point in time, there could be thousands of observations coming in that need to be updated and combined with prior models before a weather prediction could be made.
Despite the enormity of the task, Nychka seemed excited to continue refining the process of weather prediction. Maybe one day he’ll crack the case of the mysterious Utah weather.