New PR/next steps about climpred HOT 6 CLOSED

Something I noticed after merging. Although your perfect model notebook compiled fine after some edits, two figures changed. I'm not sure if this is because of a data/function change since the notebook was originally uploaded or if I messed something up.

(1) See the table under "Predictability Horizon".

My version (with updated code from PR):

Your version from before the merge:

(2) See the map under "Bootstrapped threshold":

My version:

Your version:

from climpred.

If bootstrap was set to a value less than 500, it might not give stationary thresholds. This then affects PH. I think it’s fine

from climpred.

See bradyrx/esmtools#4. We should add this load_dataset function to climpred as well. Then we can have sample datasets to load just like in xarray. We should probably rename our sample files to make them easier to recognize and load. We can make the default github_page keyword to this repo and then add an extension to our sample data folder.

from climpred.

Okay, the load_dataset feature is done without needing any new dependencies. Example:

from climpred.loadutils import open_dataset, get_datasets

get_datasets()
>>> 'MPI-DP-1D': decadal prediction ensemble area averagesof SST/SSS/AMO.
>>> 'MPI-DP-3D': decadal prediction ensemble lat/lon/timeof SST/SSS/AMO.
>>> 'MPI-control-1D': area averages for the control run ofSST/SSS.
>>> 'MPI-control-3D': lat/lon/time for the control run ofSST/SSS.
>>> 'CESM-DP': decadal prediction ensemble of global meanSSTs.
>>> 'CESM-LE': uninitialized ensemble of global mean SSTs.
>>> 'ERSST': observations of global mean SSTs.
>>> 'CESM-reference': hindcast simulation that initializesCESM-DP.

open_dataset('CESM-reference')
<xarray.Dataset>
Dimensions:     (d2: 2, time: 816)
Coordinates:
  * time        (time) datetime64[ns] 1948-01-31 1948-02-29 ... 2015-12-31
    z_t         float32 ...
Dimensions without coordinates: d2
Data variables:
    time_bound  (time, d2) float64 ...
    SST         (time) float32 ...

from climpred.

I would prefer to label my current data as perfect-model. MPI-PM-LY-1D

from climpred.

Updated in 9ff9534. I did keep "DP" instead of "LY" for consistency with CESM.

from climpred.