Dan Goodman on Nostr: Data analysis question. How can I compute my total ascent from Google fit location ...

Data analysis question. How can I compute my total ascent from Google fit location data of my run? It's quite noisy so sum(relu(diff(height))) massively overestimates. I can smooth but what's the right width? I don't know the distribution of sensor noise and it might depend on all sorts of uncontrollable factors. This feels like someone must have definitively solved this? Another option would be to project my lat/long (which seems much less noisy) onto a global height map and use that to replace Google elevation sensor, although maybe that smooths too much?