The axis of the SOFAR channel is where acoustic energy can travel without losing much intensity. Great elliptic paths shown are the shortest paths to @ctbto_alerts sensors. But sound is lazy and bends toward media with lower sound speed. What were the actual paths? #ARASanJuan 
Acoustic paths are repelled horizontally by high sound speed (warm water), shallow depths, and high latitudes. Source: Slide 22 of noiselab.ucsd.edu/view/intro2hyd…, @Scripps_Ocean
Can we guesstimate that the combined effects cancel out for the path to Crozet? Not sure about Ascension.
Can we guesstimate that the combined effects cancel out for the path to Crozet? Not sure about Ascension.
The path to Crozet actually passed underneath some sea ice at its most extreme latitudes. The sound speed profiles down there indicate that the sound may have been bouncing off the ice as it traveled.
Which would help explain why the energy received on the @ctbto_alerts hydrophones at Crozet Islands (HA04) did not show an energy spike similar to Ascension Island (HA10).
So, assuming the sound did travel along the geodesic paths from the event to hydrophones, then this is my attempt to estimate the source location using the time of arrival at Ascension H10N, Ascension H10S, and Crozet H04S. Surprisingly close to CTBTO.
Surprised at how close it turned out to matching CTBTO's original ellipse. CTBTO has access to additional data to improve their estimate like signal phase across the hydrophones and additional land based sensors. Before anyone asks, this doesn't change anything about the search!
Also, I calculated 13:51 UTC as the time of the event, matching CTBTO.
This is the combination of the hydroacoustic anomaly geolocation results and the possible areas the sub could have travelled to from the last known positions.
Update: assuming geodesic path to Crozet causes time errors of a few minutes. Due to the strong horizontal temperature gradient, the true path would have been in warmer, faster waters. So the center of my ellipse should be a bit northeast. Explains my 3 minute error from CTBTO.
Another conclusion in my analysis was that the decreasing time delays between the signals arriving at hydrophones HA10S and HA10N show that the signals were not produced from separate explosions. They are reflections of the same event off of the continental slope. #ARASanJuan
The explanation for this is in code blocks 36-44 of my notebook posted here. nbviewer.jupyter.org/github/mjc808/…
