Symbolizing geospatial data with QGIS

QGreenland Researcher Workshop 2023

Symbology is important

Symbology is critical to understanding data! Consider:

• Does your colormap look uniform across your values?
• Do your symbology choices unintentionally exclude any users, for example people with Color Vision Deficiency (CVD), a.k.a. “colorblindness”?
• What, aside from your data, will be in your visualization?
  • Basemap?
  • Other data elements?

• “Perceptual uniformity” is important for even users with perfect sight. The “Jet” colormap suffers from poor perceptual uniformity, and that creates features in the data where there are none, or hides features that would otherwise be seen.
• There are many types of Color Vision Deficiency. Look for colormaps that are friendly to all forms of CVD.
• Ensure contrast with your basemap: A light basemap requires a colormap with saturated (colorful) light values, and a dark basemap requires a colormap with saturated dark values.
• Ensure contrast with your other data elements. Red dots representing ice core locations will not show up well over temperature data displayed with the heat colormap, which is also red.

Colormaps

Colormaps are the primary way we visualize raster data. Colormaps are also useful in visualization of vector data in cases where we want to color vector features according to an attribute (e.g., magnitude of earthquakes). More on vector-specific symbology considerations in upcoming slides!

Perceptual uniformity

Jet and Rainbow are common, but “lie” about your data.

Jet

Turbo, a Google alternative to Jet

Viridis

Jet’s perceptual derivative shows high rate of change (red) some places, low rate of change others (green)

Viridis’s perceptual derivative is consistent

viscm, software used for analyzing colormap quality, was used to generate these images.

The colormaps on the left column are test images showing vertical lines that emphasize the rate of change in the colormap. If some vertical lines stand out more than others, that’s a sign that the colormap contains false features (these are areas of high perceptual derivative). If some vertical lines are invisible, that’s a sign that the colormap is hiding features of your data (these are areas of low perceptual derivative).

Jet’s lack of uniformity creates false features and hides real features in your data.

Turbo was developed by Google to replace Jet/Rainbow. It is slightly better, but not perceptually uniform.

Viridis is perceptually uniform and CVD-friendly, and was developed for the matplotlib project by Stéfan van der Walt and Nathaniel Smith (they created viscm for this purpose).

Presenter clicks the “Viridis” link (or the “for this purpose” link) and recommends reading the blog post by Stéfan and Nathaniel.

Note that viscm is currently broken but I’ve become a project maintainer and am hoping to release a fixed version in coming weeks.

Good places to start

• cpt-city: A website with lots of colormaps, some better than others.
• SVG2ColoR: A plugin available from the official QGIS plugin repository for loading SVG styles from various sources (e.g., cpt-city).
• cmocean: All perceptually uniform, some CVD-friendly.
• Colorcet: All perceptually uniform, some CVD-first.
• ColorBrewer: All perceptually uniform, most CVD-friendly.
• CMasher: All perceptually uniform, some CVD-friendly.
• matplotlib: Some perceptually uniform, some CVD-friendly.

Any of these sites are good places to start looking for the right colormap for your visualization.

ColorBrewer colormaps are research-based, of high quality, largely CVD-safe, but are built for a specific purpose: choropleth maps. They have low saturation in their light colors and high saturation in their dark colors, so they’re not well-suited for use on a light basemap.

My current favorite set of colormaps is cmocean by Kristen Thyng and others.

Reminder: CVD stands for Color Vision Deficiency!

Colormap set up in QGIS

Some high quality colormaps (e.g. Viridis, Cividis, Magma, Inferno, and Plasma), are built in, others need to be installed.

cpt-city and ColorBrewer in QGIS

Settings > Style Manager > “+” button > Catalog: cpt-city

Add a style

Follow these same steps but select the ColorBrewer catalog if you prefer those colormaps.

If you prefer to browse colormaps on the cpt-city website, you can use the “SVG2ColoR” QGIS plugin mentioned on a previous slide to load their colormap SVG files in to QGIS as color ramps.

List styles by author, then select cmocean

Adding “deep” colormap

Once the colormap is added, you’ll be able to find it when selecting a “color ramp” in the layer symbology editor.

There are other ways to access these colormaps in the QGIS interface.

Vector symbology

Colormaps are important for symbolizing vectors as well as rasters, but vector data have some other unique concerns.

Symbolizing shape

• Markers: color, shape, size
• Lines: color, thickness, pattern
• Fill: color, pattern
• Clustering: physical proximity

The Bird protected areas layers use hatch-filled polygons to indicate areas that may have restricted access

Line pattern could include dashed, dotted.

“Clustering” can be accessed with the “Point Cluster” vector symbology type.

Presenter reads the image caption.

Symbolizing attributes

• Labels: e.g. placenames
• Marker size: e.g. scalar value
• Color: e.g. scalar value
• Scale-dependent rendering

The Earthquakes layer uses size and color to indicate magnitude.

• Labels: Previous slide for example, the name of the place is an “attribute” in the vector data.
• Marker size: In the image, larger points indicate larger earthquakes.
• Color: In the image, darker red points also indicate larger earthquakes.
• Scale-dependent rendering enables you, for example, to hide labels when zoomed out too far (where the text might overlap).

QGIS symbology features are deep, and the documentation is very good. Please use the exercise time to explore!

Exporting QGIS symbology

QGIS supports importing and exporting symbology choices as QML files.

Use the “Style” dropdown in the layer’s Properties window

The “Save Layer Style” interface. Note that the “Load Style” interface looks similar.

Choose which categories you want to save to the style in the “Save Layer Style” interface. If you only want to share symbology, uncheck everything else. However, it is often helpful to share additional properties (e.g., Labels, which are also an important component of how data is visualized).

Exporting QGIS layer definition

QGIS supports importing and exporting a layer’s definition (symbology and pointer to the data ) as QLR files.

Particularly useful for layers that you want to share that are accessed via the internet (e.g., a WMS layer).

• To create a QLR file, right-click on a layer in the Layers Panel, then select “Export > Save as Layer Definition File”.

• To import a QLR file, use the Layer menu (“Layer > Add From Layer Definition File”)

This is a good slide to skip if we are short on time.

Style files (QML) can also be exported this way (“Export > Save as QGIS Layer Style File”).

Exercises

💪 Exercise: Symbolizing datasets together

20 minutes

💬 Discussion: On symbology decisions

20 minutes