improve snow height records by working from L0 data instead of fixing L1

[jasonebox]

we'll get a better snow height records when working from L0. That of course will be a lot of work. But in that effort, we can eliminate some of the noise that comes from sonic signal reflection from drift snow layer.
The use of photos and field notes should increase certainty.

[BaptisteVandecrux]

For info, when I translate you suggested changes into the L1 processing code:

1. I apply the filter (removing outliers) to the instrument heights (HW1 and HW2) from which the surface heights (HS1 and HS2) are derived.
2. I apply the shifts (up or down at a certain date) to HS1 and HS2 only. Because if the station tilts suddenly, the change in surface height needs to be corrected, but the height of the instrument are truly changed and therefore do not need adjustment.

Does that address your concern?

Photos are informative but tedious to work with.
There are a lot of notes about ice melt and snow depth at Swiss Camp that could help constrain the surface height record.