When looking at BOTH the attached flights they certainly create some question marks ... this due to the timing in them & where the motor start is located on the time line.
When it comes to the .TXT log it start to record when the motors go on ... that usually mark 0sec in the graph, as from one of my own
MA1 logs just below where I start up the motors with a CSC command.
But the second attached log here in the thread show this ... the motor start is at 252sec ?!?
Further more the log include 2 HP positions ... One at the same location as the first attached log, but that HP is updated just after take off to a location very close to the beach. This update of the HP just after take off is how it works ... the HP is always verified/updated just after take off.
The 2 flight logs attached in the thread is actually linked (they both points to the same DAT log. FLY061.DAT) as they are actually done during the same power on but the OP took the first flight & started then further away from the river, flew 252sec then landed & the motors shut down ... then took off again without powering down & flew until the connection was lost.
Comparing the different HP positions from both logs & plotting them in GE show this ... the red path with red pins from the first flight, the yellow path with yellow pins from the second take off ... note that the updated HP position for the second attached log (yellow path) aligns to where the first log (red path) ends.
Another view of this here ... have circled the first take off spot to the right & the second at the left.
Actually ... I think you started the motors 3 times during the same power on period. This due to that even the first log you attached don't start up at 0sec instead at 25sec. I think you shortly started up the motors during 25sec before the first log you attached (the red flight path in the pics.) & that even the .TXT log before the first you attached points to the same DAT log as the 2 in this thread.
So the question is ... where did you think you had the HP? Could it have been that the landing/take off spot was very, very close to the water, the HP got updated to be there with some inaccuracy ... but you thought the HP was further up land all the time.
Then when the AC was in RTH to the HP very close to the water (with some inaccuracy) you became surprised that it started to land there.
As stated earlier ... the log doesn't contain the period during the lost video feed so it's hard to say anything about exactly where the AC was when it started to descend. And most likely you had RC connection to the AC when it started to land as it listened to your pause button ... but it's odd that you couldn't gain stick control as pushing pause would stop ongoing automated flight modes ... perhaps you actually lost the connection again after pushing pause.