Closed mvieno closed 4 years ago
Hi @mvieno
I moved my post-processing and plotting to Python some time ago. Still use NCL on some scripts that do not need any maintenance, but as soon as I need to add a new feature I start on Python from scratch.
My visualization needs almost exclusively data exploration. Most of the time I use HoloViews/GeoViews to create interactive plots/maps. For a quick/simple interactive plots Altair can be a good choice, but the mapping is not well developed.
For reading observation data (csv files) there is no better than pandas, and xarrays for reading and collocate model output.
For an overview of visualization packages in Python visit PyViz, and have look at the tutorials to for examples on how to use different packages together.
Cheers, Á.
Hi @mvieno
I thinks your question was answered, so I'm closing this issue. Please reopen it if not the case.
Cheers, Á.
Hi All, this is not an issue, but more a discussion on mapping tools, and maybe of interest of other EMEP users.
The NCL (NCAR Command Language) is no longer supported and many of its features have moved to Python (PyNGL and PyNIO). Although python is a good language the fact that it is split in a half (version 2 vs version 3) is a little frustrating. In addition in the HPCs I am using uses an ad-hoc setup of Python (ad-hoc conda for example) and the virtual environment not always works as it should. (https://www.ncl.ucar.edu/open_letter_to_ncl_users.shtml)
Anyway, I am thinking to recode all my plotting scripts in R, but I am interested in what you think? R is quickly taking over in the scientific community. Also R work well in Linux and Windows (Rstudio is great).
It seems like I have three options: 1) Translate the NCL scripts in Python using the NCL python packages (I have not successfully installed PyNGL nor PyNIO in my HPC yet)
2) Starting from scratch in R
3) Starting from scratch in Python
Massimo