Plastic LiDAR

LiDAR datasets allow us to work with digital facsimiles of the earth surface and its adornments (e.g., vegetation). It’s also possible to transform the digital models back into physical form. I have been trying this with my 3D printer.

Figure 1. 3D printed relief map of the Fred study site. Ellipse is the location of 10 study plots. This is a print of the digital terrain model, or bare earth model, based on the 2019 LiDAR data from Glacier Bay. Compare Figure 7 in this previous post. The print is built by adding successive horizontal layers of plastic and the visible layers are reliable elevation contours, in this case about five foot contours.

I started printing 3D relief maps a few years ago with coarser data. There are several websites where you can download printer-ready (.stl) files of user-selected map areas in most of the US. This summer I found some coarse data for Glacier Bay and printed a five inch (13.5 cm) wide model of the bay. Acrylic artist’s paints work well to decorate the printed maps.

Figure 2. Hand-painted 3D relief map of Glacier Bay proper. 3D printers are a long way from making color prints that would eliminate the hand-painting step. This 5 x 5 inch (13.5 x 13.5 cm) print is as big a square as my old printer will print.

Last fall I learned that opentopography.org was making high resolution LiDAR data available for many areas. Glacier Bay was not yet included, but the potential for printing detailed, small-scale relief maps encouraged me to buy a larger 3D printer (20 cm x 25 cm [8 in x 10 in] build area). Even with old, coarse data a bigger printer can produce impressive results.

Figure 3. 3D printed relief map of Glacier Bay printed in four panels. This is based on low-resolution topographic data. The entire map is 19 inches (48 cm) wide. If someone sent this to you, with a complete set of paints and brushes, would you paint it?

With high resolution LiDAR data, a bigger printer can make intriguing objects. Vermont acquired good LiDAR for the entire state which is now at opentopography.org where one can select a map area (which can include parts of multiple tiles) and download various data types (e.g., bare-earth or first returns) in a few different formats (e.g., point cloud [LAS] or raster surfaces [TIN]). There are options to remove noise from the data and make other tweaks as your order is prepared. I got carried away printing areas of complex 13,000 year-old glacial geomorphology near my house.

Figure 4. Four 3D printed and hand-painted panels highlighting the glacial origin of landforms in Salisbury and Middlebury. These were intended as educational aids during field trips but serve as decoration in my kitchen between outings.

Converting the LiDAR raster file from opentopography.org to a file that a 3D printer can use requires a couple of steps. There is a plugin for QGIS (DEMto3D) which produces an stl file from a raster layer. Every 3D printer has a “slicer” program associated with it that converts the stl file to the gcode instructions for the printer.

Figure 5. 3D printed relief map (DTM or bare-earth) of an esker inside the ca. 1750 Glacier Bay terminal moraine a mile from Bartlett Lake. Richard Carstensen noticed this in the new LiDAR data and I thought it deserved more attention. The vertical elevation has been exaggerated by 2x.
Figure 6. Hand-painted 3D printed relief map of the ca. 1750 Glacier Bay terminal moraine shown in Figure 5. Colors code for geomorphic features: Orange — linear ridges assumed to be eskers, Deep Yellow — till, Brown — terminal moraine, White — kettles (all dry), Green to yellow — ca. 1750 outwash distal to the moraine (lighter colors are most recent). The panel is 9.75 inches (25 cm) wide. North is up.
Figure 7. 3D printed relief map of Bartlett Cove and Cooper’s Notch (large outwash channel to the right). The park headquarters, visitors lodge buildings, and docks have been removed (it is a digital terrain model). The panel is 9.75 inches (25 cm) wide and has no vertical exaggeration. North is to the left.
Figure 8. Hand-painted, 3D printed relief map shown in Figure 7. Colors code for geomorphic features: Grey — intertidal zone, Light Brown — glacial till, Brown — terminal moraine ridge, Orange — Beardslee Formation (old outwash from the advancing Glacier Bay glacier) that has been buckled by the advancing ice, Greens — ca. 1750 outwash surfaces eroded into the Beardslee Formation, and collapses of the steep cliffs at edges of outwash channels (I tried to make lighter greens represent more recent outwash events but got really confused and gave up), Blues — Bartlett Cove, dark/light boundary is the 20 fathom isobath. North is up.

The geomorphological interpretation represented in Figure 8 is somewhat hypothetical. I did not know those orange surfaces were not ca. 1750 outwash until I saw the new LiDAR data and heard from Streveler that there was some reason to believe that they are deformed remnants of a much older outwash surface that was deposited long before the ice reached the terminal moraine. Some of the orange terrain in Figure 8 is higher than the ca. 1750 terminal moraine which is not consistent with it being outwash deposited ca. 1750. I like the Glacio-tectonic Beardslee Formation Deformation Hypothesis and am sticking with it. I want to print a wider view showing all of the putative Beardslee Formation distal to the moraine, but those LiDAR data are not yet at opentopography.org and the USGS National Map site has been down for a week.

It is also possible to 3D print relief maps with the LiDAR first returns included. The surface of such maps would be covered with a model of the vegetation growing at the site. One of the products available from opentopography.org is a raster layer (TIN) with the first returns point cloud converted to a continuous surface. I don’t know how to convert a point cloud into a continuous surface, so thank you opentopography.org.

Figure 9. Hand-painted 3D printed model of the vegetation canopy at the Klotz Hills study site. Ellipse is the location of 10 study plots. The first returns surface does a good job representing scattered trees growing above the canopy of a shrub thicket. Trees established later on the outwash fan to the left and they are smaller and less dense than on the till surface where the study plots are. Compare with Figure 8 in this post. North is to the left.

Printing a plastic forest is tricky for a couple of reasons. The printer nozzle must skip from tree to tree extruding molten plastic for each tree but none in-between trees. This requires a perfect setting for ooze control that I have not yet found for the new printer (Figure 10). Also, the common 3D printers build up from the bottom by applying plastic in layers and have only limited capability to print shapes that quickly get wider from bottom to top. So spruce trees are easy to print but cottonwood trees, or any lollipop shape are harder. The QGIS plugin that makes the stl file might be solving this problem by making all trees look like Christmas trees. One way or another, all the LiDAR-based 3D printed forests I have made look like spruce forests even when they are maple or oak forests.

Figure 10. Left: Detail of the 3D print in Figure 9. Molten plastic oozing from the nozzle between trees leaves strands that would make the print difficult to paint. Right: Same area of the print after a quick treatment with my crème brûlée torch (I knew I would eventually find a use for it). More than half of the trees at this site are cottonwood, but they all look like spruce trees in the print.
Figure 11. Oblique view of the Klotz Hills print above. Painting some random trees darker makes it look like a mixed forest of spruce and cottonwood, which it is. Although the shapes of the trees are distorted by the 3D printing process and the color of the trees is somewhat arbitrary, the height and location of each tree are derived directly from the LiDAR data and might be a fair representation of (a scaled) reality.

The 10 inch panels have great potential as teaching tools in the field to show the geographic context of natural features. They might be most important for geology and geomorphology students where forests obscure the greater landscape view. Hand-painting the panels is a big task but could also be part of the educational process. Acrylic paints are cheap and easy to use and you can quickly paint draft versions and then paint over your mistakes until you get it right (most of the panels above have at least three different colored coats of paint). When you do get it right there is also the potential for displaying the panels as curiosities or artwork (for the more talented among us).

It takes from five to 15 hours apiece to print these panels. But to a 3D printer, time and plastic are cheap, and I have a pile of prints waiting to be painted. Let me know if you would like to help — the panels are easy to mail. Do you have a favorite place in Glacier Bay you would like to have sitting on your desk?

One thought on “Plastic LiDAR”

Leave a Reply

Your email address will not be published.