Elevation, in general, suffers from the 'coastline paradox' (https://en.m.wikipedia.org/wiki/Coastline_paradox).
That is, the result you obtain depends on the scale at which you measure.

You can easily compare two results taken with the same device/method, but not two very different sources...

---

https://besse.info/
https://www.strava.com/athletes/2012033

Does the coastline paradox apply to a road on the coastline?

Yes, different mapping services can have different resolution on their elevation data in their maps, and also with regards to how how many points they use in the calcution of elevation gain on a route.
Some mapping services also use data from user-uploaded files when estimating elevation gain. This inherits poor data quality (clogged barometric sensor from rain, poor gps-elevation-data or similar) from random users (easily seen on strava segments).

Measured elevation gain from different devices also suffer from the coastline effect due to how often they register the elevation, how precice and with how many decimals it logs the elevation, and how much smooting is used on the data from the source (barometric sensor or gps).

On my garmin edges I can observe the smoothing of elevation from the barometric sensor when doing hill-repeats and u-turning on steep gradients. The elevation difference between the lowest and highest point can be 100 metres, but the registered elevation can be for instance 96 metres because the algorythm smoothes out the elevation on the top and bottom over a few seconds. Please don't ask me about my everesting-ride where I found out about this ;-)

My 1030 has it at 4500-5000 feet. pulling data even 6 seconds I believe. (I updated this when I had a chance to look at old files). I did 4,200 feet yesterday...

---

Randy Christofferson(http://www.rcmioga.blogspot.com

Insert Doubt. Erase Hope. Crush Dreams.

How reliable is the Garmin 1030? I thought the elevation sensors on cycling computers weren't that accurate.

I've always gotten right around 4k