LiDAR Terrain Analysis for Concealment and Disturbance

An airborne LiDAR system fires laser pulses (typically near-infrared, 1064 nm) at rates of 100,000 to several million pulses per second from a scanner oscillating across the flight path. Each pulse can generate multiple returns as it passes through canopy layers before hitting the ground. A full-waveform system records the entire backscatter time series; a discrete-return system records the time of each significant peak, typically the first return (top of canopy), last return (ground surface), and up to 4-7 intermediate returns.

Range to each return is calculated as half the round-trip travel time multiplied by the speed of light. Combined with the scanner angle and the aircraft position from a high-accuracy GPS-inertial navigation system, each return becomes an XYZ point in a georeferenced coordinate system. The result is a dense three-dimensional point cloud of the surveyed terrain.

Raw LiDAR point clouds mix returns from the forest canopy, understory, buildings, and the ground surface. Ground classification algorithms use elevation, return number, and local neighbourhood geometry to assign each point to a ground or non-ground class. Three algorithms dominate forensic-relevant workflows: the Progressive Morphological Filter (PMF), the Multi-Scale Curvature Classification (MCC), and the Cloth Simulation Filter (CSF), which simulates a cloth dropped from above and identifies the points it would settle on.

Once classified, ground points are interpolated to a regular grid raster, the DTM, using inverse-distance weighting, triangulation (TIN-to-raster), or kriging. The choice of interpolation method and output grid cell size both affect the fidelity of micro-topographic features. A 0.25 m grid is the practical minimum for grave detection; a 1 m grid will miss most grave-scale anomalies.

A bare-earth DTM contains the terrain signal, but the human eye cannot detect a 15-cm elevation difference in a flat-coloured elevation raster. The standard forensic-analysis toolkit applies three derivatives that amplify micro-topographic contrast.

• Hillshade at multiple azimuths: a single hillshade illumination may miss features aligned parallel to the light direction. Running hillshades at 45-degree azimuth intervals and comparing the stack is the standard protocol. Sky-view factor and multi-directional hillshade algorithms automate this.
• Slope: the gradient surface computed from the DTM highlights edges where terrain changes abruptly. A grave pit edge, a soil-scrape boundary, and a drag mark all generate slope anomalies. Slope is less useful for low-angle terrain.
• Profile curvature: the second derivative of elevation along the slope direction. A convex mound produces positive profile curvature; a concave hollow produces negative. Both correspond to physically meaningful features. Curvature is arguably the single most sensitive indicator of grave-scale micro-topography.

Visual interpretation of these derivative layers is a skill that develops with practice. One useful protocol is to display curvature as a colour ramp (red for convex, blue for concave) transparently overlaid on a hillshade base, then scan systematically for isolated convex or concave anomalies that do not match natural landform patterns such as river channels, tree-throw pits, or terrace risers.

Once a burial site is located and excavation begins, the documentation requirement shifts from area-scale survey to sub-centimetre scene recording. Terrestrial laser scanners, instruments from manufacturers including Leica (RTC360, BLK360), FARO (Focus), and Trimble (TX series), acquire 360-degree point clouds from a fixed station at rates of millions of points per second.

A typical scene documentation workflow positions the scanner at 3-6 stations around the grave to ensure complete coverage. The individual scans are registered (aligned) using common targets or automated iterative closest point (ICP) algorithms. The merged point cloud forms a permanent, millimetre-accurate 3D record of the scene at a given excavation stage.

• Pre-excavation scan: captures the surface condition, any visible micro-topographic anomaly, vegetation disturbance, and access paths before any intervention.
• Excavation-stage scans: record each horizon as skeletal elements, artefacts, or soil features are exposed. Allow virtual sectioning through the scene after the physical record is gone.
• Post-excavation scan: documents the final state of the pit for court presentation. Combined with photogrammetric models, provides a defensible 3D exhibit.

The application of airborne LiDAR to forensic grave search has been documented in peer-reviewed literature from several jurisdictions. A foundational study by Andrew Chadwick and colleagues (published in the journal Forensic Science International) examined LiDAR detection of simulated and actual clandestine graves in UK woodland and farmland conditions. They found that 0.25-m DTMs with hillshade and curvature derivatives enabled reliable detection of grave-sized disturbances in open and lightly vegetated terrain, with detection rates falling in dense closed-canopy forest where ground-point density dropped below 4 points per square metre.

In conflict-zone and mass-grave investigations, LiDAR has been deployed by the International Commission on Missing Persons (ICMP) and similar organisations to map disturbed terrain at scales of tens of hectares. The technique identifies individual pits, vehicle track patterns, soil stockpiles, and the spatial relationship between multiple disturbances: contextual information that informs interpretation of the larger scene.

The key variables controlling LiDAR detection of a grave-scale disturbance are point density (related to flying altitude and pulse repetition rate), DTM grid resolution, and the minimum detectable elevation change given the background terrain noise. These form a linked system: high point density enables fine grid resolution, which enables detection of smaller features against a noisier background.

A soil-scrape scar, where surface organic material or topsoil has been removed to eliminate surface evidence, is among the hardest features to detect. It presents as a slight surface lowering of 2-5 cm over an area of a few square metres. Detection requires a DTM at 0.1 m resolution and highly uniform background terrain. In practice, soil-scrape scars are better detected by spectral difference in optical imagery (exposing bare mineral soil) than by LiDAR micro-topography alone.