Example for geopandas line geometry: Use Natural Earth dataset, change to Asia

[yvonnefroehlich]

Description of proposed changes

Use the Use the Natural Earth dataset in the example for geopandas line geometry. For details see #4223 (comment).
Let's hope the request / download is more stable and this will make our docs build less often fail 🚀.

For the rivers there is only one river plotted for Europe. When changing to 50 m (instead of 110 m) I get the error

File C:\ProgramData\Anaconda3\envs\pygmt_env_dev_650\Lib\site-packages\pygmt\helpers\tempfile.py:148 in tempfile_from_geojson
    for col in geojson.columns:

  File C:\ProgramData\Anaconda3\envs\pygmt_env_dev_650\Lib\site-packages\pandas\core\generic.py:6321 in __getattr__
    return object.__getattribute__(self, name)

AttributeError: 'GeoSeries' object has no attribute 'columns'

RuntimeError: GDAL Error: Features without geometry not supported by GMT writer.. Failed to write record: <fiona.model.Feature object at 0x00000206E466CFB0>. 

When using the subset rivers[rivers["scalerank"] != 5] it works. Interestingly the subset rivers_five = rivers[rivers["scalerank"] == 5] is empty. For 10 m, I do not face such an error.

Europe (currently)	Europe 110 m	Europe 50 m	Europe 10 m

Related to

• Example for geopandas polygon geometry: Use the Natural Earth dataset, change to Africa #4228
• Update the gallery example for plotting lines with LineString/MultiLineString geometry #3711
• Which data to use for geopandas.GeoDataFrame choropleth example? #2786
• Add gallery example for plotting geopandas.GeoDataFrame objects #1374

Fixes #4223

Preview: https://pygmt-dev--4229.org.readthedocs.build/en/4229/gallery/lines/linestrings.html

Guidelines

• General Guidelines for Pull Request
• Guidelines for Contributing Documentation
• Guidelines for Contributing Code

Slash Commands

You can write slash commands (/command) in the first line of a comment to perform
specific operations. Supported slash command is:

• /format: automatically format and lint the code

[michaelgrund]

Third option, just show line objects for a geopandas line geometry example, nothing else. So, no cities should appear 😉 .

[yvonnefroehlich]

Third option, just show line objects for a geopandas line geometry example, nothing else. So, no cities should appear 😉 .

Yep, here for the line example we will stay with the rivers only.
The image with the cities was from the related issue, as I was wondering if we maybe want to add a new third example for points.

[seisman]

When changing to 50 m (instead of 110 m) I get the error

File C:\ProgramData\Anaconda3\envs\pygmt_env_dev_650\Lib\site-packages\pygmt\helpers\tempfile.py:148 in tempfile_from_geojson
    for col in geojson.columns:

  File C:\ProgramData\Anaconda3\envs\pygmt_env_dev_650\Lib\site-packages\pandas\core\generic.py:6321 in __getattr__
    return object.__getattribute__(self, name)

AttributeError: 'GeoSeries' object has no attribute 'columns'

RuntimeError: GDAL Error: Features without geometry not supported by GMT writer.. Failed to write record: <fiona.model.Feature object at 0x00000206E466CFB0>. 

It's likely there are one or a few rivers that doesn't have geometry information.

This can be avoided if we focus on the rivers in Europe only. This can be done by:

rivers = rivers.cx[-10:30, 35:57]

[yvonnefroehlich]

I am wondering if it would be welcome to change the region from Europe to South America for the line geometry example. Then we can use Europe for an example on the point geometry. The plotting of the point geometry is straightforward, but I think it's good to show that users can pass the columns of a gdf directly to the Figure.text method to label features, e.g., cities, which is much easier than having to set up a GMT-specific file. I prefer to have Europe for the points example because there are no overlapping labels.

rivers, lines, South America, 50 m (this PR #4229)	cities, points, Europe, 50 m (new PR #4231)

[Esteban82]

A suggestion regarding figures in tutorials: I think that horizontally-oriented (landscape) figures are often more effective than vertical (portrait) ones. Because they don't fill the entire screen, landscape figures enable users to see the visual aid and the relevant text side-by-side, which creates a smoother and more efficient reading flow.

This approach is implemented effectively in these PyGMT tutorials:

• https://www.pygmt.org/latest/tutorials/basics/plot.html#sphx-glr-tutorials-basics-plot-py
• https://www.pygmt.org/latest/tutorials/basics/coastlines.html#sphx-glr-tutorials-basics-coastlines-py

[yvonnefroehlich]

A suggestion regarding figures in tutorials: I think that horizontally-oriented (landscape) figures are often more effective than vertical (portrait) ones. Because they don't fill the entire screen, landscape figures enable users to see the visual aid and the relevant text side-by-side, which creates a smoother and more efficient reading flow.

Hey @Esteban82. Thanks for looking at this example and your feedback. Yeah, your are right and I agree the South American map gets quite large in the gallery, which is not optimal. To get a more horizontally-orientated figure, we could use Australia or Asia as region. For Australia, there are only a few rivers for the 50 m dataset. Thus I would go with Asia.

South America	Australia	Asia

[Esteban82]

Hi Yvonne! Yes, I agree with Asia.

[seisman]

Sorry, but I prefer not to use the Asia map. China and India have unresolved boundary disputes, and from China’s perspective, the boundary lines in this dataset—and even in GMT’s GSHHG dataset—are incorrect. Since I am from China and leading this project, I prefer not to include any national boundary lines of China to avoid potential political sensitivities.

[yvonnefroehlich]

Sorry, but I prefer not to use the Asia map. China and India have unresolved boundary disputes, and from China’s perspective, the boundary lines in this dataset—and even in GMT’s GSHHG dataset—are incorrect. Since I am from China and leading this project, I prefer not to include any national boundary lines of China to avoid potential political sensitivities.

No worries! This is totally fair and an important aspect.
Would not plotting the national boundaries help? Otherwise we need another region; maybe Australia even there are only a few rivers for the 50 m dataset.

[seisman]

Would not plotting the national boundaries help?

Yes.