Add an gallery example/tutorial for interactive data visualization using Panel?

[seisman]

The EGU22 PyGMT short course provides an example showing how to use Panel for interactive data visualization with PyGMT (https://www.generic-mapping-tools.org/egu22pygmt/ecosystem.html#interactive-data-visualization-with-xarray-panel-and-pygmt).

I just tried it and found it sometimes useful to explore the effects of different GMT parameters (especially before we implement the movie method #1347).

For example, here is an example to show how the azimuth can affect grid shading (modified from https://www.pygmt.org/dev/gallery/images/grdgradient_shading.html):

import numpy as np
import panel as pn
import xarray as xr
import pygmt

pn.extension()

# Make a panel widget for controlling the depth plotted
az_slider = pn.widgets.DiscreteSlider(name='Azimuth', options=list(np.arange(0, 360, 15)), value=0)

grid = pygmt.datasets.load_earth_relief(resolution="03m", region=[35, 50, 35, 45])

@pn.depends(az=az_slider)
def view(az):
    fig = pygmt.Figure()
    pygmt.makecpt(cmap="terra", series=[-7000, 7000])
    shade = pygmt.grdgradient(grid=grid, azimuth=az, normalize="e2")
    fig.grdimage(
        grid=grid,
        shading=shade,
        projection="M15c",
        frame=["a4f2", f"+tazimuth={az}"],
        cmap=True,
    )
    fig.colorbar(frame="af")
    return fig

pn.Column(az_slider, view)

Here is a screencast showing how it works locally:

Screencast.from.2023-03-23.16-55-53.webm

Although the interactive widget doesn't work on our static HTML documentation, users can still copy-and-paste the codes to try it on their own Jupyter notebooks. So I think such a tutorial is still very useful.

[yvonnefroehlich]

I just tried it and found it sometimes useful to explore the effects of different GMT parameters (especially before we implement the movie method #1347).

Yep, this is a quite nice way to show or test the variation of one parameter or quantity, while the rest of the map / plot remains the same. Also tried this already based on this EGU ShortCourse example 🙂.

[seisman]

Ping @maxrjones for thoughts/comments on the potential gallery example using "Panel" before someone takes the time to work on it.

[maxrjones]

Yes, I think this is a good idea. I'll be happy to provide a review.

[yvonnefroehlich]

I would be happy to give this a try. @GenericMappingTools/pygmt-maintainers is there a parameter of a plotting method you have in mind to highlight in this example? Currently, I am wondering whether pygmt.meca is suitable (looking at dip, strike, or rake). Or should we better show a quantity which is not directly related to a PyGMT parameter, like in the EGU Short Course example?

[seisman]

Currently, I am wondering whether pygmt.meca is suitable (looking at dip, strike, or rake).

I'm afraid this is not a good example, because (1) three variables (strike, dip and rake) may make the example slightly more complicated (2) non-seismologists may find it hard to understand the example.

Or should we better show a quantity which is not directly related to a PyGMT parameter, like in the EGU Short Course example?

Some simpler examples in my mind:

1. A global map with varying central longitude (similar to this animation https://docs.generic-mapping-tools.org/latest/animations/anim07.html but coastlines are enough)
2. Day-night terminator line which depends on the datetime (dynamic version of https://www.pygmt.org/dev/gallery/embellishments/solar.html)

[yvonnefroehlich]

We probably have to add panel as a dependency, or? Should this be done in a seperate PR?

[seisman]

We probably have to add panel as a dependency, or? Should this be done in a seperate PR?

You can add it in the same PR as the gallery example and we will see how it works.

[yvonnefroehlich]

Currently, I am wondering whether pygmt.meca is suitable (looking at dip, strike, or rake).

I'm afraid this is not a good example, because (1) three variables (strike, dip and rake) may make the example slightly more complicated (2) non-seismologists may find it hard to understand the example.

Or should we better show a quantity which is not directly related to a PyGMT parameter, like in the EGU Short Course example?

Some simpler examples in my mind:

1. A global map with varying central longitude (similar to this animation https://docs.generic-mapping-tools.org/latest/animations/anim07.html but coastlines are enough)
2. Day-night terminator line which depends on the datetime (dynamic version of https://www.pygmt.org/dev/gallery/embellishments/solar.html)

I started working on a tutorial in PR #2498 and used the first suggestion by @seisman.