SIParCS 2021 - Jiaqi Li

Expanding and strengthening the transition from NCL to Python visualizations

Recorded Talk

In January 2019, NCAR announced plans to transition from the NCAR Command Language (NCL) to the Python scientific ecosystem. The GeoCAT team is in charge of making the transition as smooth as possible. GeoCAT-Examples and GeoCAT-Viz support the creation of a diverse set of plotting templates (GeoCAT Examples Gallery) that aim to recreate the original NCL plots almost exactly. However, Python and its related packages have limitations that result in differences between recreated plots in Python and the original NCL plots. We regard certain differences acceptable, such as tick mark formats or font sizes, as they do not deviate the plots from their intended teaching goals. We regard some differences preferable, such as choosing a better color scheme or using randomly generated data instead of hard-coded data, since they follow better practices of data visualization and better support the teaching purpose of the plot. Other differences, however, showcase limitations of Python. NCL handles and supports functionalities such as implicit axis scaling and curly vectors that cannot be exactly recreated in Python yet. Further research and investigations are needed to find solutions to these missing functionalities. Python scripts also tend to be a lot longer than their NCL counterparts. Contributions were made to both GeoCAT-Examples and GeoCAT-Viz Github repository this summer.

Mentors: Julia Kent, Orhan Eroglu, Michaela Sizemore, & Anissa Zacharias

Slides and poster