Themes

Themes can be used to customize the appearance of Sankey diagrams. sankee contains a few built-in themes, but you can also create your own themes for additional control. Let’s get started by setting up and initializing our packages.

[1]:

import sankee
import ee

ee.Initialize()

The Default Theme

sankee includes a default theme it will use unless you specify one. To demonstrate, we’ll look at urban expansion around Houston, Texas between 1986 and 2020.

[2]:

aoi = ee.Geometry.Point([-95.78867372638199, 29.773539101705214]).buffer(30_000)
sankee.datasets.LCMS_LU.sankify(
    years=[1986, 2020],
    region=aoi,
    n=100,
    title="Urban Expansion in Houston, TX",
)

Built-in Themes

You can override the default theme using the theme argument to sankify. Using theme="d3" will give us a similar appearance to a Sankey diagram from the d3-sankey library.

[3]:

sankee.datasets.LCMS_LU.sankify(
    years=[1986, 2020],
    region=aoi,
    n=100,
    title="Urban Expansion in Houston, TX",
    theme="d3"
)

Or we can use theme="simple" to get a slightly more minimalist theme.

[4]:

sankee.datasets.LCMS_LU.sankify(
    years=[1986, 2020],
    region=aoi,
    n=100,
    title="Urban Expansion in Houston, TX",
    theme="simple"
)

Building a Custom Theme

The theme argument accepts either the name of a built-in theme or a custom sankee.themes.Theme object. You can create your own Theme object to define a custom style. Let’s try that.

Link and Node Arguments

The first way we can modify a figure is by passing in keyword arguments for the link and node styles. We’ll experiment by changing the thickness and pad between our nodes and the link colors. Check out the Plotly documentation for more information.

[5]:

custom_theme = sankee.Theme(
    # We can set a custom link color to override the default. We'll choose a transparent bluish color.
    link_kwargs=dict(color="rgba(0, 180, 255, 0.1)"),
    # Thickness sets the size of the node rectangles and pad sets the spacing between them. We can
    # also customize the border around nodes using the line argument.
    node_kwargs=dict(thickness=100, pad=100, line=dict(width=4, color="#628eb3")),
)

Now we’ll apply our custom theme to our Sankey diagram.

[6]:

sankee.datasets.LCMS_LU.sankify(
    years=[1986, 2020],
    region=aoi,
    n=100,
    title="Urban Expansion in Houston, TX",
    theme=custom_theme
)

Title and Label Styles

The other way to customize a theme is by specifying CSS properties to apply to the node labels and figure title. We’ll add those to the theme parameters we defined above to finish our custom theme.

[7]:

label_style = """
    font-style: italic;
    font-size: 18px;
    color: #4e6980;
    letter-spacing: 2px;
"""

title_style = """
    font-size: 24px;
    font-weight: 900;
    font-variant: small-caps;
    color: #4e6980;
"""

custom_theme = sankee.Theme(
    link_kwargs=dict(color="rgba(0, 180, 255, 0.1)"),
    node_kwargs=dict(thickness=100, pad=100, line=dict(width=4, color="#628eb3")),
    label_style=label_style,
    title_style=title_style,
)

The final product with our custom node, link, label, and title styles:

[8]:

sankee.datasets.LCMS_LU.sankify(
    years=[1986, 2020],
    region=aoi,
    n=100,
    title="Urban Expansion in Houston, TX",
    theme=custom_theme
)