A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
From: Ian Vaughn
Date: 2017 Jul 7, 17:05 -0700
As Mike says, you've looked at this more than most. Do you know if anyone has tried a full probalistic analysis of likely crash locations? There are a bewildering number of unknowns, but I recall a suggestion in Blind Man's Bluff (Sherry Sontag & Christopher Drew, 1998, Pages 58-60, 104-106) that probabilities for some unknowns can be pinned down by asking experts to bet. The trick is not so much recreating the flight in a realistic simulator-- as good as that may look on TV-- as crudely recreating the flight 10's or 100's of thousands of times based on guesses about how a 1930's-trained navigator and pilot would behave. "Re-guessing" these behaviors with every simulation based on how, say, NavList's air navigators would bet certain things happened might help at least suggest just how absurd some of the hypotheses are. Some work would be required to figure out what questions to ask. A great deal more work would be required to figure out if the search area can be narrowed. Still, it might rule out certain scenarios "beyond a reasonable shadow of a doubt" and would be an opportunity for an interesting combination of 1930's navigation math and 2010's Bayesian modeling.
As a thought experiment, consider the following questions: in the same situation, what percentage of experienced air navigators would stay on the LOP vs. fly a search pattern vs. look for a different island? How wide a search pattern would they try? I'd be curious how TIGHAR's proposed landing sites compare with those results.
Proposed search areas are often considered proprietary by the various organizations, so I apologize in advance if this question is too close to any trade secrets.
Ian Vaughn, PhD
Deep Submergence Laboratory
Woods Hole Oceanographic Institution