search

geco-bern / FluxDataKit

LEMONTREE flux data kit
https://geco-bern.github.io/FluxDataKit
2 stars 11 forks source link

MODIS data smoothing mess #36

Closed khufkens closed 1 year ago

khufkens commented 1 year ago

Ok, I seriously doubt that the MODIS smoothing and gap filling routine makes any sense. Seriously f-cked up stuff IMO. Anyway, the current routine introduces spurious results throughout and in particular at year end changes, with sharp peaks in LAI/FPAR for anything that isn't northern hemisphere seasonal as far as I can tell.

stineb commented 1 year ago

What routine/function are you referring to? Something originally from FluxnetLSM or from ingestr?

khufkens commented 1 year ago

The cobbled together pile of scripts that was used in addition to FluxnetLSM, the PLUMBER-2 workflow (if you can call it that).

stineb commented 1 year ago

The explosion of extrapolated values is a common issue, but depends on the type of smoothing algorithm. I implemented several in ingestr. Maybe a useful alternative?

khufkens commented 1 year ago

I know how to handle these data, it's just yet another fix/failure in this whole setup.

khufkens commented 1 year ago

Fixed using the phenocamr workflow, which is more robust and parsimonious across data types, instead of the double tripple smoothing stuff in FluxnetLSM.

khufkens commented 1 year ago

Needs pull of https://github.com/computationales/FluxDataKit/tree/modis_fix