Most of what we know about the Moon is learned by looking at it. That is a problem for anyone who cannot see well, or cannot see at all. A telescope image tells a sighted student that a crater is a bowl with a wall and a bright ray system. It tells a student who is blind nothing. The idea behind this project is simple: if the shape can be felt, then the lesson can reach people who would otherwise be left out of it.
Copernicus is a good place to start. It is one of the most recognisable craters on the near side, and its shape carries most of the basic lessons about impact craters in one object: a bowl, a steep wall, terraces, a central peak, and a floor that is lower than the land around it.
Why Copernicus
The crater sits in Oceanus Procellarum, close to 9.6° N and 20.1° W. Giovanni Battista Riccioli named it in 1651 after Nicolaus Copernicus. It is roughly 93 km across. Geologists date it to about 800 million years ago, which is young by lunar standards, and it gave its name to the Copernican period, the most recent period in the lunar timescale. Its bright ejecta rays spread across the maria for hundreds of kilometres, which is why it stands out even on a small full Moon.
For a tactile prop, size and clarity matter more than anything else. Copernicus is large enough to show the full structure of a complex crater, and its features are bold rather than subtle. That makes it easier to feel than a smaller, shallower crater whose rim barely rises from the plain.
Where the shape comes from
The terrain is built from SLDEM2015, a digital elevation model of the Moon. It merges laser altimetry from NASA’s Lunar Reconnaissance Orbiter (LOLA) with stereo imagery from JAXA’s Kaguya Terrain Camera. The source has a resolution of 512 pixels per degree, about 59 m per pixel near the equator.
There are two versions of the model, and they are built at different resolutions on purpose. The CNC prototype design is a 20 cm square mesh of about one million triangles, made from the full 512 pixels-per-degree data, so the small terraces and the texture of the ejecta blanket are kept in the design. The web version is much lighter. It covers a 132 km square around the crater, resampled to a 600 × 600 grid, so each point is about 220 m across. That is coarser than the design, but it keeps the terraces, the rim and the central peak, and it keeps the page small enough to open quickly on a phone. Heights in the web version are stored as 16-bit values and packed into the page as a base64 string, so the whole model is one file with no server behind it.
Measured against the surrounding plain, the model’s lowest floor point is about 3.0 km below, its rim crest about 1.5 km above, and the central peak about 1.6 km below. The central peak therefore stands roughly 1.5 km above the floor. These are the grid’s extreme values at 220 m per point, so they are the numbers the simulation panel shows. Those are the differences a learner needs to feel, so they matter more here than they would in a photograph.
Turning heights into a model
The build is straightforward, and each step is a choice that changes how the object reads by hand:
- Heights to a mesh. A flat plane with 600 × 600 vertices, each vertex lifted to the height of its cell. Normals are recalculated so the surface shades smoothly.
- Vertical exaggeration. Real crater walls are gentle when measured against their width. Across 93 km, the rim is a slight rise that a hand would hardly register. Exaggerating the height pulls the walls, terraces and peak up into something a finger can follow. The simulation starts at 2× and runs from 1× to 6×.
- Light. A single low sun brings out relief through shadows. For a viewer who can see, a low sun makes the walls and the central peak stand out. The sun-elevation slider starts at 20°.
- Colour by height. Low ground is darker and high ground is lighter, so the same model works for a viewer who relies on contrast as well as one who relies on touch.
The exaggeration and the sun angle are the two settings that change the object most, and they are the two I tuned while building the prop. Too little exaggeration and the crater reads as a flat disc. Too much and the floor turns into a hill. The goal is the point between those two.
Designing it to be touched
This is still an ongoing project. The design has not been cut yet, and the material is still being explored. I will cover that in the next post. For now, the surface is not a direct copy of the height data. The prototype design was exported in 2×, 3× and 4× vertical exaggeration while the design is being explored, and these are the decisions in the current design:
- Smoothing. The surface is smoothed with a Gaussian filter (sigma 0.6 mm at the design scale). This removes small noise and keeps the large shapes, such as the walls, terraces and central peak, that a hand should notice first.
- Rounded edges. Corners have a 10 mm radius, and the top edges of the block are rounded too, so the object is comfortable to hold and explore without catching fingers.
- A direction marker. A half-circle groove on the north wall tells the reader which way is north, so the orientation is found by touch before the rest of the crater is explored.
Try the simulation
The model below is the same terrain as the prop. You can drag to rotate it, scroll or pinch to zoom, and use the two sliders to change the vertical exaggeration and the sun angle.
Why it matters
Telescopes, observatories and planetarium shows are built around sight. Some of the people who would enjoy the Moon most are the ones those places serve least well. A model that can be touched, turned over in the hands and explored at the pace of the person holding it gives a different kind of access. It also gives teachers something concrete to pass around the room.
I am an astronomer, not a specialist in tactile teaching, so I expect to learn a lot from the people who use these objects. This post is the first step in sharing how the model was made, and I would be glad to hear what would make it more useful.