During a 1937 attempt to circumnavigate the globe in a Lockheed Electra aircraft, Amelia Earhart and her navigator Fred Noonan disappeared over the central Pacific Ocean near Howland Island.
The nuclear submarine USS Scorpion was declared missing after failing to return to home port in 1968 as scheduled.
Finding the Scorpion
Dr. John Craven (Chief Scientist of the U.S. Navy’s Special Projects Division) lead a team to find the Scorpion. As part of their analysis, they employed something called Bayesian search theory.
The team produced a map showing the probabilities for the USS Scorpion’s most likely resting place.
The Scorpion was found, just 220 yards, from the maps indicated most likely position for the Scorpion.
Bayesian search theory
Bayesian Search Theory is based upon Bayesian Probability, which I posted about previously under Introducing the Reverend Bayes.
The steps for Bayesian Search are something like these:
- Constructing scenarios using subject matter expertise for what might have happened,
- For each scenario, construct a probability distribution for the location, based on the scenarios events,
- Combine the above scenarios information to produce an overall probability distribution for locations, this is something like a like a probability contour map,
- Search the ares of highest probability first.
Will it work for Amelia
I found the following site interesting nr16020report.org. NR16020 was the registration number of Amelia’s Lockheed Electra aircraft.
This site by George W. Anderson and Gundars Osvalds, have written a paper detailing an entire process for finding Amelia’s aircraft, including Bayesian Search.
The paper is available from their site as a PDF document. It talks about Bayesian Search in chapter 6.
Chapter 9 talks about the Bayesian Search grid and Figure 9-3 shows the final search grid with the probabilities they have come up with.
Search for Amelia’s aircraft: