With hvplot explorer soon supporting xarray/gridded datasets, I think the next logical step for hvplot explorer is a CLI.
Before moving forward with the CLI, I would like to solicit feedback from users of Panoply, ncview, and other CLI viz tools:
- What situations do you use those tools for?
- What do you like about those tools?
- What would you like to see in those tools?
- Would it be useful to wrap a CLI around hvplot (i.e. would you use it?)
To provide my thoughts as a past user of panoply:
- When I need to quickly examine a data file, e.g. model output without having to type the following (before hvplot was released):
import xarray as xr
ds = xr.open_dataset("path/to/file")["take time to figure out var name"].isel(remove_other_variables=0)
ds.plot(x="take_time_to_figure_out_x", y="take time to figure out y name")
And sometimes, I would just create a .py file / notebook in the data directory, which is a bit disorganized.
- A plot immediately shows, support for a variety of file formats,
- Interactively zoom-in/out and check for values at points
- I would
I use these tools less in my current role, but as a support scientist and model developer I used these quite a bit (and so did my peers).
What situations do you use those tools for?
- quick looks at data files (e.g. for checking model runs, data shared w/ me, checking for anomalous values or patterns, easy slices and profiles, exploration)
- tutorials (lower barrier to entry)
- visualization tool suggestion for collaborators or folks unfamiliar w/ formats common in weather / climate sciences (especially as a way to get started or for less technical folks)
What do you like about those tools?
- no configuration needed, but options available
- easy to install / often available as a module on HPC systems
- powerful way to quickly visualize and inspect data
- easy and quick
- low barrier to entry
- abstracts away a lot of the complexities of reading and visualizing complex data
What would you like to see in those tools?
- better performance
- a more modern look / UI
- more interactive visualization options / inclusion of metadata
- easy ways to overlay geospatial reference data (maybe?)
Would it be useful to wrap a CLI around hvplot (i.e. would you use it?)
- Yes, but I’m not sure it’d replace these other tools
- Might be tough to compete with some of the ease of installation / use of some of these other tools though - I get the feeling that’s a big part of what keeps them going
Hope this helps and excited to see where you go with this!
I used ncview a lot and only use it to vizualize data. I know that Panoply has more advanced features, but as a graduate student I’m just interested in understanding my outputs and ensuring they seem correct. It works on both Windows and linux systems and has several good features, but the most important is that it’s quick and easy. I think a neat feature would be for the CLI to view a grib file as well as .nc file. I haven’t seen any applications to currently do that which are easy for just anyone to download and use.
Because of the need for high-throughput data processing, I do a lot of work on the command line, and until a couple years ago had relied on ncview. However, I often work with data that is unstructured native-grid data and so decided to develop a new tool that could produce quick-look visualizations of unstructured grid data without the need for regridding. The result of these efforts, entitled “ncvis” can be found here: GitHub - SEATStandards/ncvis: NetCDF File Viewer
The tool is undergoing active development, and I’m open to feature requests.