Sixteen wavelengths in an open excavation
Researchers from Aarhus University and Moesgaard Museum have developed a multispectral imaging rig that uncovers hidden stratification directly inside active excavation trenches. The prototype illuminates vertical soil sections using sixteen distinct narrow-band light-emitting diode wavelengths, sweeping sequentially from near-ultraviolet through visible bands to near-infrared. Rather than relying on broadband white flashes or ambient sunlight, the camera records reflection and absorption profiles frame by frame across each color band.
The team tested the first iteration at Sorte Muld on the Danish island of Bornholm, a site known for complex, dark Iron Age cultural layers where successive occupations can blur together. Subsequent processing brings previously invisible layer boundaries, fill episodes, and buried ground surfaces into focus. A secondary benefit appears in the ultraviolet range, where bone fragments fluoresce, allowing excavators to locate organic materials that standard trench photography overlooks entirely. The team deployed a second-generation setup at a degraded Viking Age cemetery at Fredbjerg in northern Jutland, reducing capture times from five minutes down to thirty seconds per station.
Why photogrammetry and lidar fall short on soil
Excavation is an irreversible destructive process. Once a field crew digs through a stratum, physical relationships disappear forever. Modern field recording relies heavily on standard photogrammetry, structured light scans, and terrestrial lidar to register coordinate positions and produce high-resolution digital elevation models of surfaces. However, standard photogrammetric color balancing and lidar intensity returns struggle when soil units share identical brown or black values.
Integrating multi-band reflectance into spatial modeling solves this limitation. Instead of mapping uniform soil as flat textures, surveyors can project classified multispectral data back onto dense point clouds and textured polygon meshes. The technique delivers hyperspectral-grade spectral discrimination at roughly one-twentieth the financial cost of typical hyperspectral sensors. This brings laboratory-level techniques directly into the trench.
Where the method goes next
As the developers work toward on-site processing and mobile device integration, field crews can use machine learning to segment units and locate artifacts during the dig rather than months later in post-processing. In commercial surveying and geomorphology, mounting multi-band illumination rigs alongside camera sensors on robotic platforms or low-altitude drones will expand how specialists map exposed cliff faces, erosion gullies, and soil horizons.
Adding multispectral channels to photogrammetric processing pipelines and Gaussian splats will allow teams to build photorealistic 3D models where users can toggle spectral bands and inspect chemical or moisture variations in real time. By bringing low-cost, multi-wavelength imaging into routine documentation, the system establishes a new baseline for how field science preserves ephemeral spatial data.
Read More: https://curiosmos.com/led-imaging-helps-archaeologists-see-buried-layers-hidden-in-ordinary-soil/
