part 2: tutorials compatible to oggm_v16

[lilianschuster]

This is the second part of making the tutorials compatible to OGGM v1.6:

Apart from _use_your_owninventory.ipynb, all other notebooks that I worked on run with OGGM v16 now, but there are a few things to check.

Here is a list of changes that I did and some things that need to be checked :

preprocessing_errors

• changed URLs, basically shows now preprocessing errors of version 1.6 and 1.4 because only in 1.4, we also have different climate files at the moment
• the error flowline_model_run_historical only occurs in level 4, so after the climate was processed. I changed that accordingly
• added a rough comparison of failing glacier area between the versions run_with_a_spinup_and_gcm_data:
• changed to L5 elev_band base_url
• set y0=1965 (instead of tstar of HEF) and removed the temperature_bias=-0.5, it still creates a slightly different behavior than before, i.e. the different projections don't end perfectly at the same volume. But it is approximately ok. Otherwise,
• [ ] @anoukvlug could maybe check if the result is still as expected.

use_your_own_inventory:

• some errors that I could not solve, probably the output of utils.get_rgi_glacier_entities(['RGI60-11.00897']) changed, or the geopandas plotting stuff??
• [ ] @fmaussion, can you have a look there?

ingest_gridded_data_on_flowlines

• elev_band L3 base_url added
• obs_ice_vel_x -> itslive_vx , ...
• needed to adapt the recalibration, the OGGM_inversion_velocity is now two to three times higher
• [ ] @fmaussion could maybe check if I did something wrong there or if that is expected. Basically the last four plots are different compared to the stable version.

merge_gcm_runs_and_visualize

• elev_band L5 base url
• added a section that visualizes data with matplotlib and seaborn. I also show how to compute median, interquartile range and total range (I guess this is a good robust alternative to mean and std. for small numbers of GCMs). I did not yet find a good page where this is explained; feel free to add a good link or more explanation there
• also needed to change the min_count when doing .sum() in xarray. Otherwise (at least in my xarray version), summing up two np.NaN results in a zero (which is very bad when estimating averages or median afterwards ... )
• moved import Holoviews and import Panel to the last section like that the notebook can run until then even if the two packages are not installed
• added two options (default is to use ISIMIP3b data for projections)
• I didn't touch the Holoviews section (and somehow couldn't manage to install the package). But normally, when you use option='CMIP5', the last section should work
• [ ] @pat-schmitt could maybe check if the Holoviews/Panel section runs (you need to set option=CMIP5)

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[fmaussion]

Thank you! Included in my recent PR - leaving open for dicussion