Select to expand quote
Imax1 said..On a not fast day ,after a crash ,my GW 60 clocked 75 kts ,2 sec

When I got home and downloaded it was around 24 kts. That seemed right because my 5x10 was around 22 kts. At no stage in the crash , my wrist went anywhere near 75 kts. What actually is happening here ? What has corrected this , the watch or ka72. Sorry I'm a total numpty when it comes to this kind of stuff.
I'll attempt to answer this in very general terms.
GPS receivers (more generally, GNSS receivers) are constantly tracking the individual signals from all of the available satellites. As a consequence of signal tracking, the GNSS receiver knows a couple of useful things for each satellite:
1) The approximate distance between the satellite and the receiver. This is referred to as the pseudo-range and is based on the travel time of the signal which is travelling at the speed of light. I'll not go into atmospheric effects in this post.
2) The rate at which the satellite and the receiver are moving towards each other (or away from each other). This is based on the frequency of the received signal. The received frequency is different to the original frequency due to the Doppler effect.
Here's a nice little animation of the GPS constellation in its early years (just 21 satellites). The visible satellites are shown in red.

Position and time are determined from the pseudo-range observables (distances between the various satellites and the receiver) using a process known as trilateration. Time needs to be determined at the same time as position, but I won't go into why right now.
Speed can be determined in a similar way but it is based on the rate at which the receiver is moving towards (or away) from the individual satellites. Since this is typically performed using the Doppler observable, our community uses the colloquial name "Doppler Speed".
Lots of factors affect receiver accuracy, but during a crash the signal tracking can be severely impaired and result in "loss of lock". Under these circumstances the observables can be very wrong for several seconds, until such time as the signals are being tracked properly.
I won't go into how the signal tracking is performed in this thread. To do that topic any justice would require a lot of detail. Suffice to say during any period that signal tracking is severely impaired and the observables are very wrong, position and speed will be impacted.
KA72 corrected the data for you. It either did this using the speed accuracy estimate saved by the GW-60 (aka SDOP / SDOS), number of satellites being tracked, HDOP (which relates to the satellite geoemetry) or the acceleration being unrealistic.
Incidentally, I wrote a short series which attempts to explain the principles behind GNSS (including signal structures).
medium.com/@mikeg888/what-is-gps-gnss-fd0092a794ea