Where is the tropopause ?

[fpavogt]

Describe the bug
We need to clarify/settle on the equation/algorithm for the tropopause. Currently, dvas supports 3 algorithms (wmo1957, mch, gruan) that all follow the 1957 WMO definition, but with the latter two adding some custom tweaks to better handle high-resolution sounding profiles.

We should agree which is best, possibly devising a new (better) option, and clearly describe it in the docs.

For reference, here is the (current) description of the three options in dvas:

'wmo1957': 
"The first tropopause is defined as the lowest level at which the lapse rate decreases to 2degC/km or less, 
provided also the average lapse rate between this level and all higher levels within 2 km does not exceed 2degC/km"

'mch': 
The first tropopause is defined as the lowest level at which the lapse rate decreases to 2deg/km or less,
provided also ALL the lapse rateS between this level and all higher levels within 2km do not exceed 2deg/km.

'gruan':
The first tropopause is defined as the lowest level at which the lapse rate decreases to 2deg/km or less,
provided also ALL the MEAN lapse rateS (between this level and all higher levels, COMPUTED FROM ALL
SEQUENTIAL LEVEL PAIRS LOCATED WITHIN A GIVEN LEVEL INTERVAL) within 2 km do not exceed 2deg/km.

The gruan algorithm typically finds a tropopause level within 1s=~5m of the value quoted in the RS41 and iMS-100 GDPs. The difference may be related, at least in part, to the additional smoothing of the gph profile performed by GRUAN in their GDP assembly chain, which is not implemented in dvas.

[fpavogt]

The gruan algorithm works well, with the exception of F16, F37 and F38. These flights should be looked at more closely.

[fpavogt]

The implemented GRUAN algorithm works well. More accuracy is not required for the purpose of the UAII analysis, that deals wit large atmospheric regions (FT, UTLS, MUS). No further experiments will thus be conducted on this front.