Hi all - I posed the same question direct to Bryton Support and got this response -
"We've check the second 3 track, following is the dump of all Valley -> Peak altitude difference. The difference between these three track is coming from the different input data. Every ride for the same route will generate different (x,y,z) serial of data. We believe that you are riding on exactly the same route, the data will just not be the same.
If you have interest the following is our analysis.
The format and data as below:
<track point index>:<peak or valley>:<altitude> <alt diff>
track id 241773, 243209, 246054
4:V:39.3 -> 11

:51.5 dz=12.2, 4:V:44.8 -> 11

:57.8 dz=13.0, 9:V:49.4 -> 49

:84.2 dz=34.8
30:V:47.7 -> 58

:79.1 dz=31.4, 19:V:51.0 -> 55

:86.2 dz=35.2, 59:V:78.5 -> 69

:84.8 dz=6.3
163:V:47.9 -> 177

:56.9 dz=9.0, 155:V:49.2 -> 174

:59.3 dz=10.1, 157:V:50.2 -> 169

:59.4 dz=9.2
198:V:56.9 -> 261

:83.0 dz=26.1, 193:V:58.9 -> 247

:81.8 dz=22.9, 193:V:59.9 -> 245

:82.2 dz=22.3
269:V:77.4 -> 329

:147.3 dz=69.9, 253:V:74.5 -> 309

:146.6 dz=72.1, 260:V:80.1 -> 316

:149.0 dz=68.9
.... skip
1499:V:50.6 -> 1510

:55.6 dz=5.0, 1569:V:53.8 -> 1581

:58.8 dz=5.0, 1483:V:47.5 -> 1491

:52.5 dz=5.0
gain 461.2, 515.8, 449.9
About the first 2 tracks issue. because we used to change the algorithm but not re-calculate all old tracks. So if you can ride same route should get close altitude gain.
Best regards,
Bryton support team."