torresr-cnrm / xoce

Python tools for analysing and offline diagnosing ocean and climate datasets

XOCE - Xarray based Ocean and Climate Experiments

XOCE is a suite of Python codes and tools for analysing and offline diagnosing global ocean datasets.

---

NOTE

The library is still in development and code is probably not free of bugs.

---

xoce's Issues

[API] improve the experiment class

Manage the processing outputs

• processing could be define in the experiment class, for instance average/2D map/profiles/etc..
• allows saving an experiment object in a JSON ou HDF file where a link to the processing file are stored (useful for loading directly the processing files whitout recomputed average each time)
• allows loading an experiment only by its name

Note: this should be think about a NEMO run folder
run_name
output
var1.nc
var2.nc
post
average
maps
experiment1.ext
experiment2.ext

Boundaries management

Need to automatically compute some boundary cells (depth_bnds, lat_bnds, e3t, e1t, etc..). Also, when interpolation is need or if a coordinates is changed (to re interpolate on another grid) we need to change the bnds also.

Check if functionality exists with xarray ?
Check how Antares (cerfacs) manage this ?

Create Installation section in README.md

Warning when installing cartopy, if issue is found, use the solution find here : https://stackoverflow.com/questions/67518451/accept-use-of-deprecated-proj-api-h-error-when-installing-cartopy

export CFLAGS="-I/usr/include -DACCEPT_USE_OF_DEPRECATED_PROJ_API_H=1"

Setup simple tests

Create tests for validating library. Some ideas :

• API test
• calculator test
• io test
• etc..

Add automatically attributes in dataset resulting from a processing object

After using a processing (treatment) of XOCE, it should be good to write information in the datasets generated :

• the old attributes of the input datas (need to mark it as "origin")
• the parameters of the treatment use and its values
• the version of XOCE

Attributes and units management

In some processings, data manipulation involve new variable units or attributes. For instance :

• Integration is sometimes weighted by volume cell and should change variable units
• In a general way, processing could automatically replace some classical attributes (min/max, date, etc.) and add some nemopy info (version, post-processing date, etc.)

processing dataset results attributes

Nice improvement : each processing should define a function for writting attributes in the resulting datasets.

Informations to have :

• xoce version
• processing parameters
• description of the processing

Improve Reader/Writer interface

• First, manage dataset and arrays coordinates
• Then, another improvement should be to create a real API with class Experiment in order to store post-processing datasets in a class which allows easily (re-)processing and plotting (instead of a list of xarray datasets)
• Add metadata in writers : at least XOCE version and post-processing date ; maybe link to the post-processing file ?

[CMIPExperiment] Remove double longitude points when loading

Figure and plotting API

Create classes for plotting graphes + some functions

• carto

  • colorbar access (range, colormap, normalization, etc.)
  • projection (cartopy)
  • x, y axes
  • size, title, etc.

• 2D graphes (zonal and meridional average)

• 1D or profiles or 0D time evolution (volume average)

• statistics (get plot from Adele internship)