Aerial and Satellite Remote Sensing

Before the satellite era, the most systematic aerial coverage of the UK and much of Western Europe came from RAF photographic reconnaissance during the Second World War. These prints, held by Historic Environment Scotland's National Collection of Aerial Photography (NCAP) and by national archives in other countries, provide an exceptionally detailed pre-disturbance baseline. Any feature visible in the current terrain but absent in a 1944 image has formed in the last eight decades: a powerful constraint on what a change can represent.

Post-war national mapping programmes, agricultural surveys, and planning applications have generated further photographic coverage at regular intervals. In England, the Historic England aerial photograph archive holds millions of images. In the US, USGS Earth Explorer provides free access to historical aerial photography back to the 1930s for much of the country. Google Earth Pro's historical layer provides global coverage with frequent revisits from the mid-2000s onward, and its time-slider function is a standard DBA tool.

• Soil marks: bare ploughed soil shows disturbed areas as colour contrasts when the subsoil has been mixed into the topsoil by digging.
• Crop marks: differential growth in cereals, grassland stress patterns, and shadow marks over grass are most visible from low-level oblique photography in dry summers.
• Shadow marks: low sun angles emphasise micro-topographic anomalies such as grave mounding or subsidence hollows.
• Multi-temporal comparison: a disturbance present in imagery from 2008 but absent in 2005 constrains when the ground was broken, potentially corroborating intelligence about timing.

Human decomposition releases nitrogen, phosphorus, and a range of organic compounds into the soil column above a buried body. These nutrients alter the growth of overlying vegetation in ways that are detectable in the near-infrared and red reflectance bands long before they are visible to the naked eye. The NDVI is the most widely used measure, but it is not the only one.

Field experiments published in peer-reviewed literature, including work by groups at Cranfield University and the University of Adelaide, have confirmed that decomposition-driven vegetation anomalies are detectable with multispectral sensors in both temperate and arid environments. The anomaly is strongest in the first two growing seasons after burial and can persist for years if the soil chemistry remains altered. It is, however, non-specific: nutrient enrichment from animal carcasses, farm waste, or fertiliser application produces identical spectral signatures.

This non-specificity is why spectral anomaly detection is a triage tool, not a confirmation. Every NDVI anomaly requires ground-based follow-up before it can be interpreted forensically.

Uncrewed aerial vehicles (UAVs, commonly called drones) have transformed the practicality of remote sensing in forensic search. A fixed-wing platform carrying a multispectral sensor can cover 50 to 100 hectares per flight; a multirotor operates at lower altitudes in constrained spaces and provides higher spatial resolution. Both produce overlapping imagery that a Structure from Motion (SfM) workflow converts to a georeferenced orthomosaic and digital surface model.

1. Flight planning
   Mission planning software (DJI Pilot, QGroundControl, Pix4Dcapture) defines the flight path, overlap percentage (typically 75-85% front, 65-75% side), altitude, and image trigger interval. Airspace clearance is obtained in advance.
2. Ground control points
   Surveyed GCPs: markers placed at known coordinates using RTK-GPS before the flight: are essential for sub-decimetre accuracy. Without GCPs, a UAV orthomosaic has a relative accuracy of several tens of centimetres, insufficient for court use as a precise spatial exhibit.
3. SfM processing
   Agisoft Metashape, OpenDroneMap, and similar software align the overlapping images, build a dense 3D point cloud, and generate the orthomosaic and DSM. Processing time for a 200-image survey is one to three hours on a capable workstation.
4. Anomaly marking and reporting
   The georeferenced orthomosaic is imported into GIS. Spectral anomalies are marked, their grid coordinates extracted, and a field investigation priority list generated. The processing workflow is documented for chain-of-custody purposes.

Remote sensing has been applied in environments where ground investigators could not safely operate. The International Criminal Tribunal for the former Yugoslavia (ICTY) made extensive use of commercial satellite imagery to document mass graves, vehicle movements, and burial-site changes in Bosnia during the mid-1990s. The imagery was used as direct evidence in the Krstic and Blagojevic trials, establishing the spatial and temporal relationship between satellite-observed disturbances and witness testimony about executions.

Open-source satellite programmes have significantly expanded what non-governmental investigators can access. UNOSAT (the UN's satellite analysis centre) has published imagery analysis of suspected mass graves in Syria, Iraq, and Myanmar, providing evidentiary input to international accountability processes. Maxar Technologies' open data programme released high-resolution imagery of conflict sites in Ukraine from 2022 onward, enabling independent verification of reported atrocities. The Sentinel-2 constellation (ESA) provides free 10-metre multispectral imagery with a five-day revisit cycle, sufficient for change-detection analysis at regional scale.

The chain of custody for satellite imagery in legal proceedings requires documentation of the acquisition date, sensor specifications, any processing applied, the identity of the analyst, and the methodology of interpretation. Raw pixel data is provided to the defence alongside any interpreted product.

Remote sensing methods share a common limitation: they detect surface or near-surface signatures. They cannot confirm what is beneath the anomaly. A strong NDVI signal is consistent with a buried body but also consistent with a dozen other explanations. Dense canopy cover prevents aerial sensors from seeing the ground at all. Urban environments are cluttered with spectral noise. Deep burials in heavy clay may not produce any surface signature for years.

• Canopy cover: closed woodland severely limits optical and multispectral sensor performance. LiDAR, which penetrates canopy, is the primary remote-sensing tool in forested terrain.
• Time since burial: fresh disturbance has a stronger soil-mark signal than a well-settled grave several years old. Crop marks are seasonal and depend on the growth stage and weather conditions at acquisition.
• Non-unique anomalies: NDVI enrichment and soil marks are produced by animal carcasses, sewage seepage, collapsed drains, and agricultural practices as readily as by human burials.
• Cost and access: high-resolution commercial satellite tasking has a lead time and a cost. In time-critical investigations, the acquisition window may have passed before tasking can be arranged.

The practical integration model is therefore layered: satellite change detection defines the regional-scale candidate area, UAV multispectral survey maps anomalies within it at higher resolution, canine and geophysical survey ground-truths the highest-priority anomalies, and targeted excavation confirms or eliminates each. Remote sensing compresses the funnel; ground methods close it.