DOV-Vlaanderen / groundwater-logger-validation

Analysis on validation methods for groundwater logger data
MIT License
2 stars 2 forks source link

Drifts 34/35: detect_drift() v02 #70

Open DavorJ opened 4 years ago

DavorJ commented 4 years ago

This is the continuation of v01 (#67), but here we use multiple reference timeseries. For reference timeseries the new KNMI data is used (f6a2173). Here it is visualized (the lines are transparent and blend):

knmi_all

It shows that the KNMI timeseries are almost identical, but not all are available before august 2008.

Here is the output.

And here is the comparison with #67. The results are very similar. Due to differences in resulting model variance (and sometimes pre 2008 data), the significance is sometimes different by a factor of 10. For example:

image

image

Dip is a bit less pronounced.

image

More data taken into consideration and the drift of -0.09 cmH2O/year is less significant...

image

And here is the output in CSV format. All output details of this analysis are here.