Arctic Sea Ice

Rationale

The datasets are large, complex. Navigating them can be challenging, especially for those without specialized training or tools to interpret the data effectively.

While traditional visualizations like 2D graphs and charts are helpful, they don't fully convey the spatial nature of Arctic sea ice data.

My goal is to experiment with a more immersive, 3D visualization—almost like a game—to make this data more intuitive and accessible.

Technical Process

The source dataset consists of two key elements: gradient-based images representing ice distribution, and numerical values for ice extent and volume. However, these elements are stored in multiple folders and files, making comparisons time-consuming.

To streamline this, I transformed the data so it could be read quickly in React. As users scroll, the relevant satellite image updates instantanousely, allowing quick interrogation of large volumes of data.

2D to 3D transition

In Blender, I model a sphere, import the sea ice extent texture, then use a shrinkwrap modifier to wrap it tight against the surface of the sphere. The is used an input to a custom shader on the globe model. Using the legend as a height look-up-table, I displace the height along the globe's normal vector, thus giving a 3D representation for sea ice concentration.

The camera zoom is controlled by scrolling and when the maximum zoom is reached it transitions to a top down view of the satellite image.

Three things are controlled:

• 1. Camera position
• 2. Camera zoom
• 3. Appearance of the plane (that renders the satellite image)

Are the ice caps shrinking?

Let's now try and answer some questions using the data visualisation. I added some filters for yearly minimum and maximums. We can see that indeed, the yearly minimum and maximum (the month in which the sea ice is smallest and greatest) appear to trend downwards.