Subplots

When you’re preparing a figure for a paper, there will often be times when you’ll need to put many individual plots into one large figure, and label them ‘abcd’. These individual plots are called subplots.

There are two main ways to create subplots in GMT:

• Use pygmt.Figure.shift_origin to manually move each individual plot to the right position.

• Use pygmt.subplots to define the layout of the subplots.

The first method is easier to use and should handle simple cases involving a couple of subplots. For more advanced subplot layouts however, we recommend the use of pygmt.subplots which offers finer grained control, and this is what the tutorial below will cover.

Let’s start by importing the PyGMT library

import pygmt

Define subplot layout

The pygmt.subplots command is used to setup the layout, size, and other attributes of the figure. It divides the whole canvas into regular grid areas with n rows and m columns. Each grid area can contain an individual subplot. For example:

fig, axs = pygmt.subplots(nrows=2, ncols=3, figsize=("15c", "6c"), frame="lrtb")

will define our figure to have a 2 row and 3 column grid layout. figsize=("15c", "6c") defines the overall size of the figure to be 15cm wide by 6cm high. Using frame="lrtb" allows us to customize the map frame for all subplots instead of setting them individually. The figure layout will look like the following:

for index in axs.flatten():
    i = index // axs.shape[1]  # row
    j = index % axs.shape[1]  # column
    fig.sca(ax=axs[i, j])  # sets the current Axes
    fig.text(
        position="MC", text=f"index: {index}, row: {i}, col: {j}", region=[0, 1, 0, 1]
    )
fig.end_subplot()
fig.show()

Out:

<IPython.core.display.Image object>

The fig.sca command activates a specified subplot, and all subsequent plotting commands will take place in that subplot. This is similar to matplotlib’s plt.sca method. In order to specify a subplot, you will need to provide the identifier for that subplot via the ax argument. This can be found in the axs variable referenced by the row and col number.

Note

The row and column numbering starts from 0. So for a subplot layout with N rows and M columns, row numbers will go from 0 to N-1, and column numbers will go from 0 to M-1.

For example, to activate the subplot on the top right corner (index: 2) at row=0 and col=2, so that all subsequent plotting commands happen there, you can use the following command:

fig.sca(ax=axs[0, 2])

Finally, remember to use fig.end_subplot() to exit the subplot mode.

fig.end_subplot()

Making your first subplot

Next, let’s use what we learned above to make a 2 row by 2 column subplot figure. We’ll also pick up on some new parameters to configure our subplot.

fig, axs = pygmt.subplots(
    nrows=2,
    ncols=2,
    figsize=("15c", "6c"),
    autolabel=True,
    margins=["0.1c", "0.2c"],
    title="My Subplot Heading",
)
fig.basemap(region=[0, 10, 0, 10], projection="X?", frame=["af", "WSne"], ax=axs[0, 0])
fig.basemap(region=[0, 20, 0, 10], projection="X?", frame=["af", "WSne"], ax=axs[0, 1])
fig.basemap(region=[0, 10, 0, 20], projection="X?", frame=["af", "WSne"], ax=axs[1, 0])
fig.basemap(region=[0, 20, 0, 20], projection="X?", frame=["af", "WSne"], ax=axs[1, 1])
fig.end_subplot()
fig.show()

Out:

<IPython.core.display.Image object>

In this example, we define a 2-row, 2-column (2x2) subplot layout using pygmt.subplots. The overall figure dimensions is set to be 15cm wide and 6cm high (figsize=["15c", "6c"]). In addition, we used some optional parameters to fine tune some details of the figure creation:

• autolabel=True: Each subplot is automatically labelled abcd

• margins=["0.1c", "0.2c"]: adjusts the space between adjacent subplots. In this case, it is set as 0.1 cm in the X direction and 0.2 cm in the Y direction.

• title="My Subplot Heading": adds a title on top of the whole figure.

Notice that each subplot was set to use a linear projection "X?". Usually, we need to specify the width and height of the map frame, but it is also possible to use a question mark "?" to let GMT decide automatically on what is the most appropriate width/height for the each subplot’s map frame.

Tip

In the above example, we used the following commands to activate the four subplots explicitly one after another:

fig.basemap(..., ax=axs[0, 0])
fig.basemap(..., ax=axs[0, 1])
fig.basemap(..., ax=axs[1, 0])
fig.basemap(..., ax=axs[1, 1])

In fact, we can just use fig.basemap(..., ax=True) without specifying any subplot index number, and GMT will automatically activate the next subplot.

Shared X and Y axis labels

In the example above with the four subplots, the two subplots for each row have the same Y-axis range, and the two subplots for each column have the same X-axis range. You can use the layout option to set a common X and/or Y axis between subplots.

fig, axs = pygmt.subplots(
    nrows=2,
    ncols=2,
    figsize=("15c", "6c"),
    autolabel=True,
    margins=["0.3c", "0.2c"],
    title="My Subplot Heading",
    layout=["Rl", "Cb"],
    frame="WSrt",
)
fig.basemap(region=[0, 10, 0, 10], projection="X?", ax=True)
fig.basemap(region=[0, 20, 0, 10], projection="X?", ax=True)
fig.basemap(region=[0, 10, 0, 20], projection="X?", ax=True)
fig.basemap(region=[0, 20, 0, 20], projection="X?", ax=True)
fig.end_subplot()
fig.show()

Out:

<IPython.core.display.Image object>

Rl indicates that subplots within a Row will share the y-axis, and only the left axis is displayed. Cb indicates that subplots in a column will share the x-axis, and only the bottom axis is displayed.

Of course, instead of using the layout option, you can also set a different frame for each subplot to control the axis properties individually for each subplot.

Advanced subplot layouts

Nested subplot are currently not supported. If you want to create more complex subplot layouts, some manual adjustments are needed.

The following example draws three subplots in a 2-row, 2-column layout, with the first subplot occupying the first row.

fig, axs = pygmt.subplots(nrows=2, ncols=2, figsize=("15c", "6c"), autolabel=True)
fig.basemap(
    region=[0, 10, 0, 10], projection="X15c/3c", frame=["af", "WSne"], ax=axs[0, 0]
)
fig.text(text="TEXT", x=5, y=5, projection="X15c/3c")
fig.basemap(region=[0, 5, 0, 5], projection="X?", frame=["af", "WSne"], ax=axs[1, 0])
fig.basemap(region=[0, 5, 0, 5], projection="X?", frame=["af", "WSne"], ax=axs[1, 1])
fig.end_subplot()
fig.show()

Out:

<IPython.core.display.Image object>

When drawing the three basemaps, the last two basemaps use projection="X?", so GMT will automatically determine the size of the subplot according to the size of the subplot area. In order for the first subplot to fill up the entire top row space, we use manually adjusted the subplot width to 15cm using projection="X15c/3c".

Note

There are bugs that have not been fixed in the above example.

In subplot mode, the size of each subgraph is controlled by the figsize option of pygmt.subplots. Users can override this and use``projection`` to specify the size of an individual subplot, but this size will not be remembered. If the next command does not specify projection, the default size of the subplot mode will be used, and the resulting plot will be inccorect.

The current workaround is to use the same projection option in all commands for the subplot. For example, we forced subplot (a) to have a different size using projection="15c/3c. The next command within the subplot (e.g. text) must also use projection="x15c/3c", otherwise the placement will be wrong.

Since we skipped the second subplot, the auto label function will name the three subplots as a, c and d, which is not what we want, so we have to use fig.sca(A=""(a)" to manually set the subplot label.