Following the post yesterday on the new satellite elevation graph, I have modelled a course with constant rate of turn that ends at the final search box, and which fits the elevation graph. As stated yesterday, a great circle course at constant speed leads to the original search area.
In this work I fitted the start point (lat, long), original heading, change of heading per hour and the (constant) speed, and of course the satellite elevation data. I have fitted the 19:41 to 00:11UT elevation data. The fit is shown below.
Best Fit To The Satellite Elevations Released 29-Apr-14 And The Final Search Box Photo by RichardC10 | Photobucket
Some notes:
1. The best fit was a change of -10degrees/hour, from a start heading of 194.3, speed of 327kt.
2. The yellow point is the fitted course extended back to 18.27UT (the time of the pings previous to 19.41) so this point is not a fit to any data. The green box is the last reported radar position at 18:22UT. The model interpretation only requires that MH370 was established on the white arc of the track before the 19.41UT ping time. The exact route from the green box to the interception point on the white arc is not determined but I have plotted a small sector of the 18:27UT ping arc to show this is consistent.
3. The fit to the elevation data is excellent, r.m.s. error of 0.1degree, which is as good as the data can be read from the published fuzzy image (this may be a sign that the data has been over-fitted – the number of fitted parameters is large).
4. The plotted course is not consistent with an interpretation of the published BFO/Doppler data developed by myself and others. However, Inmarsat have said that their BFO model was ‘fine-tuned’ so it is quite possible (indeed almost certain) that the BFO data changed from the published set.
5. Constant magnetic heading courses to the final search box do not give enough heading change to fit the elevation data.
This graph does not prove that MH370 went to the area of the final search box. Presumably, the authorities used the results of modelling later generations of the BFO and elevation data, and perhaps other data sources. I think the graph does indicate that there is at least one such course (there will be others) that is consistent with the elevation data and hints at the route taken.