I am a satellite communications engineer and just want to give a bit of insight on the ping data that was used to draw the two corridors.
First of all I want to comment that the ping data can provide a pretty accurate (within a few km) picture of the range of the target to the satellite. In that respect the target can be within a circle that has as a centre the sub satellite point. Someone also mentioned about multiple satellites just to remind that the inmarsat system uses geostationary satellites and there is just one that covers that part of the globe otherwise known as IOR (Indian Ocean Region)
The result corridors are not an outcome of just ping ranging but I would assume that they are correlated with other data. What I would have done is start from the last point of radar contact and draw a circle that would define all the possible locations of the aircraft at the time of the next ping. That circle would intersect to at least at one point the ranging circle see
here that are defined by the next ping roundtrip delay. Then repeat the same for every ping I got and should end up with some locations across the final arc. Each one of them can have an associated probability by correlating the probabilities of altitude and ground speed of the plane.
I therefore see no problem at the depicted arcs, although I am sure certain locations or sub arcs have higher probabilities than the others. Good luck to them this is a very tough case.