On 20/02/2018 12:43, N_Cook wrote:
My Jason1 +Â* Jason2 + Jason3 concattenated graphical plot
http://diverse.4mg.com/jason1+2+3r.jpg
2003 to Nov 2017
retaining as much as possible of the 3 separate images, 2mm
discontinuities, linear "fit" segments etc.

Other such long-term plots

http://www.kpress.info/images/Jan_20...level_rise.jpg

https://www.aviso.altimetry.fr/filea... rence_sm.png

Anyone doubt a curve is a better fit than linear, then just a matter of
what sort of curve is the optimal fit.

A curve might well be a (slightly) better fit, but I am not convinced
the data are strong enough to merit fitting an extra free parameter.

http://diverse.4mg.com/jason1+2+3r.jpg

I am curious as to why Jason1 data is smooth but Jason2 has a distinct
sinusoidal variation with a 3 year period. There is some weak evidence
for a change in slope but it is just within the noise. More recent data
seeming to have quite a lot more of it than earlier in the decade.

--
Regards,
Martin Brown