Common Terrain Print Failures

First: is it the data, the model or the print?

Open the model in the slicer and look at it with smooth shading before printing again. If the defect is visible on the model, no print setting will fix it; go back to the data or the mesh. If the model looks right and the print does not, it is a print problem.

Stair-steps and terraces on gentle slopes

What you see: flat shelves with sharp edges across hillsides, like contour lines.

Cause 1, layer height (on the print only). On a gentle slope each layer steps a long way sideways. Step width is layer height divided by the tangent of the slope angle, so 0.2 mm layers on a 5° slope step about 2.3 mm Formula. Fixes, in order: variable layer height, a thinner layer height, then more exaggeration if the model allows it. Details in orientation and layer height for terrain.

Cause 2, the heightmap (visible on the model too). An 8-bit grayscale heightmap holds only 256 levels, so a big elevation range is stored in coarse steps that the mesh reproduces as terraces Formula. If the terraces appear in the slicer's shaded view, rebuild the model from the original DEM or a 16-bit heightmap. See DEMs and heightmaps.

Holes, pits and spikes

What you see: deep square pits, flat-bottomed holes, needle spikes or blocks that are not landforms.

• Data voids. Some datasets have cells with no measurement. USGS says the SRTM 1 arc-second global data were void-filled by NGA using interpolation with other elevation sources, but that "some tiles may still contain voids." Missing cells are stored with a nodata value; if a converter reads that value as a height instead of as missing, it becomes a deep pit or a spike.
• Filling voids. GDAL's gdal_fillnodata fills nodata regions by interpolating from valid pixels around their edges. Its -md option sets how far, in pixels, it searches (default 100 Maker spec), so large gaps may not fill well; -si adds smoothing passes, which changes the filled values. A filled area is an estimate, smoother than real ground.
• Surface models. The Copernicus DEM is a surface model that includes buildings, infrastructure and vegetation. Towns and forests print as blocks and bumps, and exaggeration makes them worse. Use a bare-earth dataset where one exists for your area, or accept them as part of the surface.

Fix all of these in the data or the mesh. The slicer cannot tell a data error from a cliff.

Thin or cracked base

What you see: a crack, a see-through spot or a soft, flexing area, usually along a river or valley floor.

Cause: the base is measured from the bottom to the lowest terrain point, and that is where the print is thinnest. On a tile, a valley running edge to edge is a natural fold line.

Fix: increase base thickness under the lowest point Measure it, check there are enough bottom and top solid layers in the thin region in the slicer preview, and shift tile cuts so they do not run along a valley. See bases and thickness.

Warped corners on big flat bases (FDM)

What you see: corners lift from the bed, tiles rock on a table, or seams between tiles show a step.

Bambu Lab explains warping as fast cooling of printed layers creating stress between them. Prusa notes it is most common with high-temperature materials such as ASA, ABS and PC blends, and that drafts from an open window or air conditioning make it worse. Bambu Lab adds that warping can happen even with PLA and PETG in a room under 20 °C Rule of thumb. A wide terrain tile is a lot of flat contact area that is shrinking as it cools.

Fixes, roughly in the order to try them:

1. Bed and first layer. Correct bed temperature for the material and plate type (Bambu Lab lists the wrong plate type selection as a common cause), a clean, grease-free surface, and a properly calibrated first layer. See first layer not sticking.
2. Brim or corner pads. Prusa recommends a brim, or a tiny cylinder a few layers tall at problem corners.
3. Cooling, by material. Makers differ here. For PLA and PETG, Prusa says those materials need a lot of cooling, and when corners lift, to check that the print fan is enabled and working and try increasing its speed, while warning that too much fan can lift the whole object. Bambu Lab recommends no part cooling for the first few layers on common filaments and lowering auxiliary fan speed if models warp. Follow your printer maker's guidance for your material and machine.
4. Infill. Bambu Lab says large models with high infill density warp more and suggests gyroid over straight-line patterns and avoiding infill above 50% Starting point.
5. Environment. Block drafts. For ASA, ABS and similar, Prusa suggests an enclosure; it notes PLA and PETG generally want strong part cooling instead.

Resin prints fail differently on large flat areas; see resin layer separation and resin orientation and supports. The general FDM lesson is Corners Curling Up Off the Bed (Warping).

Tiles that do not line up

If the edges match in the slicer but not on the table, check for a warped tile or connectors that print slightly oversized. Pins and dowels only line up when the parts are dimensionally accurate; plain straight cuts are more forgiving of slight warping (splitting large terrain). Reprint the worst tile after fixing warping, or switch to butt joints on a backing board.