There are several problems with your reasoning, actually:
* First, the distance between observation point 1 and target point is _not_ known to any accuracy, nor (more importantly) is the bearing. This eliminates observation point 1 from use in triangulation and probably eliminates the use of triangulation in and of itself.

* Second, if the observations are accurate enough, even reasonably close observation points can still be sufficient in triangulation. They aren't at all accurate.

* Third, there _isn't_ an observation point 2, unless I missed something somewhere (which is quite possible, I admit).

* Fourth, the target point may not actually be Kei's current location. We know someone grabbed her and got her out there. That doesn't mean she's still there. For that matter, unless observation points are measured simultaneously _or_ accounting for distance (which we don't have, I'll remind you), the value for the target point received from triangulation is incorrect.

As far as I can tell, trying to determine her location at this point by that method is futile because there just isn't enough information to work with.