Seeing Through the Earth: How Ground Penetrating Radar Rewrote the Rules of Excavation

 

Archaeology Today — Methods & Technology

Archaeology Today

The Journal of Record for Field Archaeology and Remote Sensing

Volume XLVII  |  Issue 2  |  March–April 2026  |  The Ground Penetrating Radar Revolution

BOTTOM LINE UP FRONT: Ground Penetrating Radar (GPR) has transformed archaeology from a discipline defined by controlled destruction into one capable of reading the subsurface in three dimensions without disturbing a single artifact. By emitting electromagnetic pulses into the ground and mapping the reflected signals, GPR produces depth-resolved images of buried walls, floors, voids, burials, pipes, and soil disturbances to accuracies of centimeters — and does so non-destructively. The technology is now standard practice across excavation planning, heritage preservation, forensic archaeology, and conflict-zone documentation. Its most societally consequential application since 2021 has been in the search for unmarked graves at former Indigenous residential schools in Canada — with dozens of confirmed surveys now reporting hundreds to thousands of "burial features" at individual sites, documented under formal community-led protocols and underpinning both the reconciliation process and legal proceedings. Simultaneously, GPR surveys of Roman cities, Egyptian pyramid complexes, English medieval landscapes, and pre-Columbian sites worldwide are rewriting architectural history at sites that will likely never be fully excavated. The combination of GPR with drone deployment, AI-assisted signal processing, and integration with LiDAR and magnetometry is producing archaeological intelligence at a pace and resolution unimaginable two decades ago.

■ Methods & Technology Report

A technology that sends radar waves into the ground and maps what bounces back has fundamentally altered what it means to excavate — and is now doing work that reaches far beyond archaeology into justice, reconciliation, and the rights of the dead.

Methods Correspondent  ·  March 2026  ·  With Reporting From Canada, Egypt, Italy, Austria, The United Kingdom, And The United States

·      3d Gpr Is The Only Near-Surface Geophysical Method That Provides True Three-Dimensional Depth Information Non-Destructively

·      5 Ha/Day Survey Rate Achieved By Multi-Channel Tractor-Mounted Gpr Array Systems At Carnuntum (Lbi Archpro)

·      8×4×2 cm Spatial Sampling Resolution Of The Malå Mira Array System Used At The Roman City Of Carnuntum, Austria

Conventional archaeology is, at its most fundamental level, a science of controlled destruction. The moment a trowel breaks the soil, a stratigraphic context is permanently altered. An excavated burial cannot be unexcavated. A wall trench cleared for sampling has been disturbed. For two centuries this irreversible character of fieldwork was simply accepted as the price of knowledge — dig carefully, document everything, and accept that what you learn is purchased at the cost of what you leave behind for future archaeologists with better methods.

Ground Penetrating Radar has not abolished that bargain, but it has dramatically deferred it. By sending short pulses of electromagnetic energy into the ground and recording the reflections that bounce back from buried features — walls, floors, voids, graves, pipes, soil disturbances — GPR allows archaeologists to map the subsurface in three dimensions before a single shovelful of earth is turned. The information arrives as a data volume: a three-dimensional image of what lies beneath, from which horizontal "time slices" at any depth can be extracted to produce layered maps of buried structures as complete, in the best conditions, as anything obtainable by excavation.

The technology's archaeological applications span from the prosaic to the profound. At the prosaic end: GPR is now standard practice for planning excavations at construction sites, guiding archaeologists to the most productive locations and steering them around sensitive areas. At the profound end: the same instrument that helps surveyors avoid cutting a Roman water pipe is the instrument that has, in the past four years, located the suspected graves of hundreds of Indigenous children at former residential schools across Canada — providing physical evidence for what survivors have long maintained, and doing so without disturbing the remains of the dead.

How GPR Works

A GPR system transmits short bursts of high-frequency electromagnetic energy — typically between 100 MHz and 2.5 GHz — into the ground through a surface-coupled antenna. When a pulse encounters a boundary between materials of different electrical properties (a wall vs. surrounding soil, an air void vs. solid rock, a grave shaft vs. undisturbed sediment), part of the energy is reflected back to a receiving antenna at the surface.

The system records the two-way travel time of each reflected pulse — measured in nanoseconds — and, knowing the approximate speed of electromagnetic waves in the local soil, calculates the depth of the reflecting feature. Hundreds or thousands of parallel survey lines, collected in a grid pattern, produce a three-dimensional volume of reflection data. Time-slice processing extracts horizontal maps at any chosen depth, creating a layered archaeological plan of the subsurface.

Key advantage: Unlike magnetometry or resistivity, GPR provides true depth information, making it the only near-surface geophysical method capable of constructing a genuine 3D subsurface image.

Key limitation: Performance degrades sharply in conductive (clay-rich or wet) soils, which absorb rather than reflect the signal. Maximum penetration depth in ideal conditions (dry sandy soil) can reach 10–20 metres at lower frequencies; in challenging urban soils, effective depth may be only 1–2 metres.

Antenna Frequency and Depth

100–200 MHz: Deepest penetration (up to 20m in ideal soil); lowest resolution. Used for deep structural surveys, large voids, pharaonic tomb shafts.

250–500 MHz: Standard archaeological range — 2–5m depth. Detects grave shafts, foundation walls, floor surfaces, pits. Used for most residential school grave surveys (250–500 MHz).

800 MHz–2.5 GHz: Highest resolution, shallow penetration (0.3–1m). Reveals individual masonry courses, tile floors, fine stratigraphic layers.

Five Decades of Improvement

From Oscilloscope Traces to Medical CT: The Technology Arc, 1974–2026

The ground penetrating radar that archaeologists use today bears little resemblance in capability to the instrument that was first deployed on an archaeological site. From 1974 to 1976, a team at Chaco Canyon in New Mexico conducted the first known archaeological GPR survey — an experiment that was, by modern standards, almost entirely blind. The equipment was bulky, expensive, and unreliable. Output was analog: voltage traces on audio tape, resembling a seismograph readout or an EKG, from which a skilled interpreter might infer the presence of a subsurface anomaly. There was no digital processing, no depth calibration, no 3D reconstruction. A buried wall might show up as a slight thickening of the trace. Whether the archaeologist was right about its location could only be confirmed by digging.

The history of GPR improvement since then is best understood as a story told in four overlapping chapters, each driven by a different technology revolution: the digitization of data in the 1990s, the development of 3D processing and time-slice visualization in the early 2000s, the emergence of multi-channel array systems in the 2010s, and the current convergence of full-waveform inversion, AI-assisted processing, and drone deployment in the 2020s. Each chapter did not merely improve on the previous one — it changed what questions archaeologists could ask.

The commercial GPR market developed rapidly in the late 1970s and 1980s, when the first affordable systems appeared. These instruments were still analog, producing single-line vertical profile traces (now called B-scans) that required expert interpretation and extensive fieldwork to assemble into a coherent picture of a site. The first archaeological applications in Europe confirmed, at least, that the instrument could detect buried walls when conditions were favorable — dry, sandy, or gravelly soil; stone or brick construction; clear dielectric contrast between the structure and its matrix. Cyprus, Italy, and sites in the American Southwest provided the early proof-of-concept cases. But the technology remained slow, expensive, and interpretively demanding.

The transformative shift came in the early 1990s with digital recording systems. For the first time, data could be stored as numbers rather than voltage curves, enabling post-processing: filtering, gain corrections, migration algorithms that collapsed the characteristic hyperbolae of point reflectors into sharp points at their true locations, and topographic adjustment for uneven ground. Most importantly, digitization enabled the development of horizontal time-slice processing — the ability to extract, from a grid of survey lines, a horizontal "depth slice" showing the plan of buried features at any chosen depth. Lawrence Conyers, one of the field's pioneering practitioners, developed much of the analytical framework for time-slice interpretation through field work in the 1990s, initially converting analog tape data into digital form by hand. By the late 1990s, commercial software had automated the process. Suddenly, GPR data produced not just a cross-section but a map — a genuine overhead view of buried architecture at a specified depth. This was the invention that made GPR archaeologically useful in the modern sense.

Through the 2000s, improved analogue-to-digital converters accelerated data acquisition speed, higher-capacity storage expanded what surveys could cover, and 3D migration techniques transformed the geometric accuracy of subsurface images. Surveys that had previously produced ambiguous smears in the data now resolved into crisp architectural outlines. The landmark Carnuntum surveys, beginning with geomagnetic work in the 1990s and incorporating GPR in the early 2000s, demonstrated for the first time that an entire Roman city could be mapped non-invasively at a level of detail sufficient for genuine architectural interpretation. Door thresholds, column bases, hypocaust floor supports, and individual water pipes became legible features in GPR depth slices — not because the physics of the instrument had changed, but because the processing had finally caught up to the instrument's inherent capability.

The Four Generations of Archaeological GPR

Generation 1 (1970s–80s): Analog single-channel. Voltage traces on audio tape. Output was a single B-scan cross-section. Required specialist interpretation. Coverage: tens of metres per day. First use at Chaco Canyon, 1974–76; first European application at Hala Sultan Tekke, Cyprus, 1980.

Generation 2 (1990s): Digital single-channel. Digital storage enabled post-processing — background removal, frequency filtering, gain correction. Time-slice generation made horizontal plan views possible for the first time. Coverage: hundreds of metres per day. Goodman's time-slice methodology, published 1994–95, became standard practice.

Generation 3 (2000s–2010s): Multi-channel arrays. Systems with 4–16 parallel channels, tractor- or vehicle-mounted, achieved survey rates of 1–5 hectares per day. Spatial sampling of 8–10 cm cross-line became achievable. RTK-GNSS positioning provided centimetre-accurate georeferencing. The LBI ArchPro MALÅ MIRA system (16 channels, 8×4×2 cm resolution) represents the peak of this generation.

Generation 4 (2020s–present): FWI + AI + drones. Full-waveform inversion reconstructs not just geometry but the electrical properties of buried materials. Deep learning accelerates signal interpretation and feature classification. Drone mounting opens inaccessible terrain. Stepped-frequency continuous-wave (SFCW) systems achieve wide-bandwidth imaging that partially overcomes the penetration/resolution trade-off.

The multi-channel array revolution of the 2010s addressed GPR's most persistent practical limitation: speed. A single-channel instrument, pushed manually across a site in a grid pattern at 25-centimetre line spacing, can survey perhaps a few hundred square metres per day under good conditions. The multi-element antenna arrays developed in the early 2000s — of which the 3d-Radar GeoScope (a stepped-frequency continuous-wave system covering 100 MHz to 2 GHz across up to 63 channels) and later the MALÅ MIRA (a 16-channel 400 MHz pulsed array) are the most archaeologically significant — shrank the practical unit of measurement from metres to centimetres while expanding daily coverage from hundreds of square metres to multiple hectares. A critical paper on the GeoScope published in 2016 noted that the then-new ground-coupled array provided cross-line spacing below 0.1 metres — fine enough, for the first time, to image features as small as 25-centimetre-wide Iron Age post-holes or the individual brick pillars of Roman floor-heating systems. These features had always been within the physical resolution limits of high-frequency GPR; what had been missing was the sampling density to reveal them across large areas.

The stepped-frequency continuous-wave (SFCW) architecture used in some of these systems also represents a fundamental departure from the traditional impulse-radar approach. Rather than firing a short time-domain pulse, SFCW systems sweep through a bandwidth of frequencies — in the GeoScope's case, 100 MHz to 2 GHz — and reconstruct the time-domain response via inverse Fourier transform. The key archaeological advantage: SFCW achieves a large bandwidth pushed toward the lower end of the spectrum, partially overcoming the resolution–penetration trade-off inherent in impulse systems of fixed frequency. A well-configured SFCW array can simultaneously image fine near-surface detail (through the high-frequency components) and deeper architectural features (through the low-frequency components) in a single pass.

The most recent and technically ambitious development is full-waveform inversion (FWI). Standard GPR data processing extracts geometric information — the shape, size, and depth of reflecting features — but throws away much of the physical information carried in the waveform. FWI, borrowed conceptually from exploration seismology, uses the complete waveform information to reconstruct not just the geometry of subsurface structures but their actual electromagnetic properties — dielectric permittivity and electrical conductivity. Published FWI results for archaeological applications, including a 2023 paper in the Geophysical Journal International demonstrating frequency-dependent GPR FWI at a resolution of approximately 0.25 metres — four times finer than comparable shallow-seismic FWI — represent the current frontier. The technique cannot yet operate in real-time in the field, but deep-learning implementations published in 2024 by Chinese and German research groups are substantially reducing computational overhead, and real-time FWI in field conditions is now a realistic near-term prospect.

"The considerable development of the GPR hardware, data processing and visualization methods over the past 30 years have opened new possibilities for the use of the GPR method."

— Trinks, Fischer et al., on re-surveying Hala Sultan Tekke, Cyprus, originally surveyed in 1980

The cumulative effect of these generations of improvement can be made concrete with a single comparison. The 1980 GPR survey of Hala Sultan Tekke, Cyprus — described as "one of the very first applications of a ground penetrating radar system in European archaeology" — was conducted with an analog instrument producing line traces from which wall reflections could be inferred but not mapped. When the same site was resurveyed in 2010 and 2012 with a modern multi-channel digital system, the result was a detailed three-dimensional plan of Bronze Age architectural remains across the entire investigated area, with individual wall thicknesses and room configurations clearly legible. The physics of the electromagnetic interaction with the buried stone walls was identical in both surveys. What had changed was everything else: the number of parallel acquisition channels, the digital sampling resolution, the positioning accuracy, the processing pipeline, the visualization software, and the interpretive framework. Where the 1980 survey confirmed the presence of walls, the 2012 survey mapped a city.

This arc from confirmation to mapping has been the defining trajectory of GPR archaeology for five decades. The question is no longer whether something is there — experienced GPR practitioners expect to find buried structure at most historically occupied sites in favorable soil conditions. The question is now how precisely it can be characterised, how quickly it can be covered, and how accurately its properties can be quantified. All three parameters continue to improve, and the rate of improvement is currently accelerating.

Roman Cities

Virtual Excavation: Mapping Complete Roman Cities Without a Trowel

Perhaps no single application of GPR has more dramatically demonstrated the technology's potential to transform archaeological knowledge than the complete survey of intact Roman towns. Our understanding of Roman urbanism has historically rested on evidence from a handful of exhaustively excavated sites — Pompeii and Ostia above all — which are exceptional both in their preservation and in the specific circumstances of their burial. These sites are, archaeologically speaking, outliers. The vast majority of Roman towns across the empire were never buried by volcanic ash; they were simply abandoned, slowly robbed of their building materials, and plowed into their surrounding fields. Their plans survive nowhere above ground, but they survive intact below.

In 2020, a Cambridge/Ghent University team published the results of the first high-resolution GPR survey of a complete Roman town in Antiquity: Falerii Novi, a 30-hectare Etruscan-foundation city in Lazio, Italy, that was abandoned in the 3rd century CE. The survey revealed, for the first time without a single trench being opened, the town's complete street grid, its forum, a rectangular temple near the south gate, an elaborate public bath complex and market building, at least 60 large private houses, and a public monument near the north gate entirely unlike anything known from other Roman cities — a colonnaded portico on three sides enclosing a 40×90-metre open square. The city also contained a network of water pipes running not just along streets but beneath individual city blocks, evidence of coordinated infrastructure planning that overturned previous assumptions about Roman urban water supply. The paper concluded that this approach "has the potential to revolutionise archaeological studies of urban sites."

At Carnuntum in Austria — the former provincial capital of Upper Pannonia and legionary base for the Roman campaigns beyond the Danube — the Ludwig Boltzmann Institute for Archaeological Prospection and Virtual Archaeology (LBI ArchPro) has spent over a decade building a definitive GPR-based model of what was once a city of 50,000 people, now lying almost entirely below an agricultural landscape. The institute uses a MALÅ Imaging Radar Array (MIRA) system — a box of 16 GPR channels mounted on a tractor — that achieves spatial sampling resolution of 8×4×2 centimeters across survey areas of up to five hectares per day. The resulting three-dimensional data volumes are comparable, in their analytical precision, to the output of a medical CT scanner.

The Carnuntum dataset has revealed not only the full plan of the civil town's forum, its temples, mansions, and commercial districts, but specific architectural details: door thresholds, column bases, floor screeds, underfloor heating systems (hypocausts), water pipes, sewers, and individual sarcophagi. The institute famously reconstructed from GPR data the complete plan of a previously unknown gladiator school — a ludus — that proved to be one of the largest training facilities in the Roman Empire, comparable in scale to the ludus magnus in Rome. The structure was subsequently reconstructed digitally and became the centerpiece of an internationally acclaimed exhibition at the Carnuntum Archaeological Park, all without a single authorizing excavation.

"GPR data acquired at such high resolution provides a three-dimensional survey result that is comparable to medical computer tomography."

— Ludwig Boltzmann Institute for Archaeological Prospection and Virtual Archaeology, on the Carnuntum MIRA system

Egypt and the Pyramids

Inside the Pyramids: Non-Invasive Imaging of Ancient Egypt's Greatest Structures

Egypt presents GPR with some of its greatest challenges and some of its most consequential applications. The dense limestone masonry of pyramid interiors and the electrically conductive clay-rich soils of the Nile Delta limit conventional GPR penetration, demanding either lower-frequency systems or complementary methods such as Electrical Resistivity Tomography (ERT) and, most recently, muon radiography. The multi-method ScanPyramids project, which has been investigating the internal structure of the Giza pyramids since 2015, represents the most sophisticated application of non-invasive imaging to a major archaeological monument.

The project's signature achievement came in 2023 with the announced detection — confirmed by GPR in combination with muon tomography — of a previously hidden 9-metre-long corridor within the Great Pyramid of Khufu, located just below the pyramid's North Face chevron stones. A follow-up 2025 paper in Scientific Reports used three-dimensional ERT to confirm and characterize the corridor's dimensions (approximately 2.5 × 2.5 metres, beginning at approximately 1 metre depth), cross-validating the GPR and ultrasound tomography results.

In the adjacent Western Cemetery — the sprawling mastaba field north of the Great Pyramid where officials and family members of Khufu were buried — a joint team from Higashi Nippon International University, Tohoku University, and Egypt's National Research Institute of Astronomy and Geophysics conducted GPR and ERT surveys between 2021 and 2023. Published in Archaeological Prospection in 2024, the results identified a previously unexamined anomaly: an L-shaped shallow structure approximately 10×10 metres, apparently backfilled with sand (suggesting an intentionally sealed entrance), above a deeper resistivity anomaly of equal area that may represent a buried chamber or tomb. The finding is awaiting Egyptian antiquities approval for follow-up investigation.

In November 2025, a study published in NDT & E International reported GPR and ERT results from the ScanPyramids team's investigation of the Menkaure Pyramid — the smallest of the three main Giza pyramids. The surveys identified two air-filled anomalies behind the pyramid's distinctive polished granite eastern façade, an area that had long puzzled Egyptologists because the finish was inconsistent with the expected construction sequence. The anomalies, the authors concluded, are potentially consistent with a second entrance corridor, giving partial physical support to a hypothesis first proposed in 2019. The team acknowledged the limitation of penetration depth and recommended muon imaging or endoscopy for confirmation.

In March 2025, a separate team announced claimed discoveries of massive underground structures beneath the Giza Plateau using satellite-based Synthetic Aperture Radar (SAR) Doppler tomography — a technique distinct from conventional GPR. The announcement attracted significant media attention and some political amplification in the United States. However, Egyptologist Zahi Hawass and other experts quickly noted that standard GPR "can scan limited depths, not exceeding tens of meters at best," and that the claimed structures (at depths of hundreds of meters) could not have been detected by the announced methods. The claims remain unverified and have not been published in peer-reviewed form as of March 2026.

■ Methodological Note: Verified vs. Claimed

The Giza SAR/Doppler claims illustrate an important distinction that readers of archaeological remote sensing reports should maintain: the difference between peer-reviewed GPR survey results (Falerii Novi, Carnuntum, the ScanPyramids NFC corridor, the Western Cemetery anomaly, Menkaure) and press-release announcements of unverified claims using novel or contested methods.

Standard archaeological GPR has well-characterized depth and resolution limits. Claims of radar-detected structures at depths of hundreds of meters in dense limestone exceed the physical limits of conventional GPR by one to two orders of magnitude and require extraordinary independent verification before they can be treated as confirmed discoveries.

Global Survey Portfolio

A Global Technology: Select Landmark GPR Surveys

United Kingdom

Stonehenge Hidden Landscapes

A pan-European scientific consortium used GPR and other techniques over four years to survey Stonehenge and its environs, discovering over 17 previously unknown monuments including burial mounds, pits, and segmented ditches within the World Heritage site. The project, part of the Stonehenge Hidden Landscapes Project, produced the most detailed subsurface map of a UNESCO site ever created without a single excavation.

Italy — Pompeii

Regio III Survey

GPR surveys of unexcavated sections of Pompeii's Regio III — areas buried under up to 8 metres of volcanic deposits — have revealed elite residential architecture in areas expected to be non-elite suburbs, including monumental fountains, apses, and domus compounds, rewriting the social geography of the city's eastern districts. The technique allows planning of any future excavations and protects what may never be disturbed.

Egypt — Bahariya Oasis

Valley of the Golden Mummies

GPR guided excavation of additional tomb groups at Bahariya Oasis following the 1999 discovery of 105 gilded Roman-period mummies. GPR surveys using a 200 MHz SIR-2000 system identified cavities in the resistive desert soil corresponding to additional tomb clusters. Subsequent excavations confirmed a strong spatial correlation between GPR anomalies and burial chambers, with many tombs containing elaborately decorated mummies now considered the finest Roman-period burials found in Egypt.

Serbia — Belgrade

Württemberg-Stambol Gate

Urban GPR surveys with dual 200/400 MHz antennas revealed the foundation remains of an 18th-century Ottoman gate beneath a modern Belgrade square. Three-dimensional data models precisely determined the geometry, size, and layout of buried columns and construction elements before excavation confirmed the results. The case illustrates GPR's particular value in densely built urban environments where subsurface complexity — pipes, cables, rubble, tree roots — would otherwise render survey ambiguous.

New Mexico, USA

Mimbres Culture Heritage Site

Northrop Grumman's Cultural SITEs initiative deployed drone-mounted GPR and magnetometry at the remote Mimbres Culture Heritage Site in Grant County, where challenging terrain had blocked traditional survey for decades. Drone-based GPR penetrated uneven ground inaccessible on foot, revealing subsurface features consistent with architectural remains of the Mimbres people, an ancient Native American community. The project established a practical benchmark for drone GPR at archaeologically sensitive sites in difficult terrain.

Israel — Hulata

Solar Panel Site Survey

A 2024 study published in Applied Sciences demonstrated effective integration of ground-based and drone-based GPR at a Bronze Age site threatened by solar panel construction, even in challenging clayey soils that typically limit GPR performance. The drone-based system detected scattered artifacts buried 1–1.5 metres below surface across the construction footprint, allowing rapid pre-construction heritage assessment. The study showed that drone GPR, properly calibrated, can overcome the traditional limitations of aerial GPR in high-conductivity soils.

Canada — Residential Schools

The Heaviest Work: GPR and the Search for Canada's Missing Children

No application of ground penetrating radar in the history of the technology has carried heavier human weight than the ongoing search for the unmarked graves of Indigenous children at former residential schools across Canada. Beginning with the Tk'emlúps te Secwépemc Nation's May 2021 announcement of 215 suspected burial sites at the former Kamloops Indian Residential School in British Columbia — findings identified by GPR survey — the technology became simultaneously a scientific instrument, a tool of reconciliation, and a flashpoint in one of Canada's most fractious public debates.

Between 1870 and 1996, more than 150,000 First Nations, Métis, and Inuit children were forced to attend government-funded, church-run residential schools. The Truth and Reconciliation Commission, which concluded its work in 2015, estimated that approximately 6,000 children died in the system — though experts note the actual number is likely significantly higher, as records were incomplete and deaths often went unregistered. The TRC's Calls to Action 71–76 specifically called for the identification of unmarked graves, the provision of information to families, and the commemoration of those buried at or near school sites.

GPR is the primary instrument through which communities are now answering those calls. Dr. Kisha Supernant, director of the University of Alberta's Institute of Prairie and Indigenous Archaeology and a member of the federal government's national advisory committee on residential school searches, has described the technical process clearly: the instrument detects soil disturbances — not bones or bodies — that are consistent with grave shafts. Survey teams look for features of the right shape, depth, and orientation, often in east-west rows (consistent with Christian burial practice), and compare them against survivor testimony and historical records to assign confidence levels ("possible" or "probable"). Most teams survey with 250–500 MHz antennas, which provide 2–3 metres of effective depth while maintaining the resolution needed to distinguish individual grave-sized features.

The volume of confirmed findings since 2021 is substantial. In January 2025, a survey of the former McIntosh Indian Residential School in the Kenora district of northwestern Ontario — guided by elders and survivors from Grassy Narrows First Nation who identified the areas of greatest concern — found 114 "unmarked burial features" consistent with graves, with over 70 consistent with possible child burials. Of these, 106 were found within the historical cemetery boundary and 8 elsewhere on the property. Technical lead Aaron Mior stated: "We're very, very confident of what these locations represent, and that's why we're not using the word anomaly. We're not using the word potential." Records from the school document at least 165 people buried there by school officials.

Earlier, in January 2023, the Wauzhushk Onigum Nation in Ontario reported 171 "plausible burials" at the former St. Mary's Indian Residential School. The Grouard Mission survey in Alberta, led by Dr. Supernant's institute, identified 169 potential graves at the former St. Bernard's Residential School. The shíshálh Nation in British Columbia announced 81 total unmarked graves at the former St. Augustine's site, including 40 announced in 2023 and 41 additional finds reported in August 2025. The Ahousaht First Nation in April 2024 released Phase 1 findings identifying potential unmarked grave locations near both Ahousaht Indian Residential School and Christie Indian Residential School.

The community-led, survivor-guided nature of the GPR work at these sites is central to its legitimacy and its meaning. The Musqueam Nation and the University of British Columbia, which have been partnering on GPR surveys since 2006, have become national leaders in developing First Nations-led protocols for this work — training Indigenous GPR practitioners, ensuring communities control both the process and the data, and insisting that outside-for-profit service providers do not "fly in, collect data, and fly away to create their reports with little guidance from the First Nation's community." This model stands in sharp contrast to the competitive, outsourced approach that characterized early years of the search.

The Office of the Independent Special Interlocutor (OSI), established by the Canadian federal government to address the issue, released its final report — titled Upholding Sacred Obligations — in November 2024, documenting the findings of ongoing searches and making recommendations for the future governance of the search process, including calls for standardized protocols and secure funding. The International Commission on Missing Persons issued interim recommendations in March 2024 on best practices for the searches, drawing on its experience with mass grave investigations from conflict zones in the former Yugoslavia and elsewhere.

NAGPRA: The U.S. Legal Parallel

In the United States, the comparable legal framework is the Native American Graves Protection and Repatriation Act (NAGPRA), enacted in 1990. The law requires federally funded museums and institutions to inventory Native American human remains and cultural items and return them to affiliated tribes.

Thirty-four years after enactment, compliance remains incomplete. Only 48% of reported human remains had been returned as of 2025. ProPublica's ongoing "Repatriation Project" investigation identified that more than 100,000 Native American individuals' remains are still held by American museums and institutions.

Revised NAGPRA regulations taking effect January 12, 2024 strengthened tribal authority, eliminated the "culturally unidentifiable" loophole that allowed institutions to defer repatriation, and imposed stricter deadlines. GPR surveys of potential burial sites on federal and tribal lands under NAGPRA's "planned excavation" provisions now routinely employ the technology to identify features requiring consultation before any ground disturbance occurs.

2024 was the third-highest year for NAGPRA repatriations on record (ProPublica database), with more than 10,300 ancestors returned to tribes — a direct result of the regulatory pressure and the increased attention generated by the Canadian searches.

The Canadian residential school searches have also surfaced a contentious public debate about what GPR findings mean and do not mean. A well-funded counter-narrative, articulated most prominently in a 2023 collection published by the Fraser Institute, argues that because GPR detects soil anomalies rather than confirmed human remains, and because most anomaly sites have not been excavated, the searches represent a misrepresentation of the actual evidence. This argument has been rejected by the majority of the archaeological and Indigenous communities involved. GPR does not claim to confirm human remains; it identifies subsurface features consistent with graves. The decision about whether to excavate belongs to the communities, not to outside critics, and several communities have explicitly decided — for cultural and spiritual reasons — not to excavate at all, viewing the disturbance of the dead as itself a harm.

Where excavations have occurred, results have been mixed. At Pine Creek in Manitoba, excavation of a church basement in August 2023 found no human remains. At other sites, confirmation is pending. The scientific consensus is that GPR is the appropriate first-step tool: it is non-invasive, culturally respectful, and provides the evidence base that communities need to make informed decisions about next steps. It does not claim more than it delivers.

Frontiers of the Field

Where GPR Is Going: Drones, Arrays, and Artificial Intelligence

Drone-mounted systems

The marriage of GPR with drone technology is resolving two of the instrument's most persistent limitations: logistical access and terrain adaptability. Traditional ground-coupled GPR requires a surveyor to push or pull the antenna across the target surface at consistent height and speed — a demanding and sometimes impossible requirement on steep slopes, swampy ground, active construction sites, or remote landscapes. Drone-mounted GPR systems, tested at the Mimbres site in New Mexico and validated in challenging clayey soils at the Hulata site in Israel (2024), achieve acceptable signal performance while opening terrain that would otherwise be inaccessible. Multi-sensor drone platforms combining GPR with LiDAR, magnetometry, and photogrammetry in a single flight represent the current frontier, producing correlated geophysical and topographic datasets in a single survey pass.

Multi-channel array systems

The MALÅ MIRA system used at Carnuntum — with 16 parallel channels on a single antenna array — represents the high-performance end of current GPR technology. Similar multi-channel systems can survey five or more hectares per day at centimeter resolution, transforming large-scale urban archaeology from a decades-long project into a matter of weeks. In the context of development-led archaeology across Europe and North America, where construction timelines demand rapid pre-excavation assessment of areas covering many hectares, multi-channel arrays have become the technology of choice for commercial cultural resource management, generating data volumes that require AI-assisted processing to analyze in reasonable timeframes. The integration of machine learning with GPR signal processing — already demonstrated in prototype systems — is expected to become standard within the present decade.

AI signal processing

The volume of data produced by high-density GPR surveys — three-dimensional reflectivity volumes containing millions of individual measurements — already exceeds what can reasonably be analyzed by human interpreters within the timeframes that development-led archaeology demands. Machine learning approaches to GPR hyperbola recognition (the characteristic shape of a point reflector in GPR data), anomaly classification, and automatic feature extraction are all active research areas. A 2023 paper in Scientific Reports demonstrated a GPR-specific convolutional neural network for burial detection in Mesopotamian contexts. Shallow machine learning methods tested on Roman town GPR data have shown promise for semantic segmentation of architectural features. As training datasets accumulate — particularly from the extensive Canadian residential school surveys — AI-assisted GPR interpretation is likely to become transformative for cemetery and burial-site work specifically.

— ✦ —

The trajectory of ground penetrating radar in archaeology runs from a specialist curiosity in the 1970s to an indispensable standard tool, and the pace of change is accelerating. The instrument now sits at an intersection of concerns that extend well beyond professional archaeology: Indigenous rights and repatriation, heritage preservation in development-threatened urban environments, the ethics of excavation in sacred landscapes, and the accountability of institutions for historical harms. It does not answer the question of whether to dig — that judgment remains irreducibly human, political, and cultural. But it ensures that the decision is made on the best possible knowledge of what lies beneath, rather than in the ignorance that once made every excavation a gamble with the irreplaceable past.

As Dr. Supernant put it to the Canadian Press: "Technology can be a very helpful tool for communities, but we don't need ground-penetrating radar." What GPR provides is evidence — non-destructive, spatially precise, depth-resolved evidence — that communities, families, and legal systems can weigh. In the hands of communities and archaeologists working together, that is a profound gift. In the wrong hands, it remains merely data.

■ Verified Sources & Formal Citations

1. Verdonck, L. et al. (2020). "Ground-penetrating radar survey at Falerii Novi: a new approach to the study of Roman cities." Antiquity, 94(375), 705–723. CBC News coverage: https://www.cbc.ca/news/world/ground-penetrating-radar-roman-city-1.5603939
2. Ludwig Boltzmann Institute for Archaeological Prospection and Virtual Archaeology (LBI ArchPro). Carnuntum GPR survey description — MALÅ MIRA system, 5 ha/day, 8×4×2 cm resolution, gladiator school discovery. Official site: https://lbi-archpro.org/cs/carnuntum/gpr_en.html
3. Neubauer, W. et al. (2014). "The discovery of the school of gladiators at Carnuntum, Austria." Antiquity, 88. Cited in: CNR Publications overview of GPR at Carnuntum and Pompeii. CNR Microwave Tomography GPR review (PDF)
4. Goodman, D. et al. (2011). "Ground-penetrating Radar in the Regio III (Pompeii, Italy): Archaeological Evidence." Regio III elite district discovery. Published in: Near Surface Geophysics. Abstract: https://ui.adsabs.harvard.edu/abs/2011ArchP..18..187B/abstract
5. Tondi, R. et al. (2019). "Mapping the undiscovered ruins of Pompeii using ground penetrating radar." 8-metre volcanic deposit penetration, elite off-centre district identification. ResearchGate summary: https://www.researchgate.net/publication/232814924
6. Molist, M. & Casini, R. (2020). GPR and microwave tomography at Pompeii's Roman amphitheater. CNR Publications. CNR paper (PDF)
7. Tejedor-Blanco, D. et al. (2020). "Using Ground-Penetrating Radar on Archaeological Sites — Revealed Hidden Archaeology." Remote Sensing. GSSI technical overview including Carnuntum application. https://www.geophysical.com/using-ground-penetrating-radar-archaeological-sites
8. Frid, M. & Frid, V. (2024). "A Case Study of the Integration of Ground-Based and Drone-Based Ground-Penetrating Radar (GPR) for an Archaeological Survey in Hulata (Israel)." Applied Sciences, 14(10), 4280. https://www.mdpi.com/2076-3417/14/10/4280
9. Nakamura, K. et al. (2024). "GPR and ERT Exploration in the Western Cemetery in Giza, Egypt." Archaeological Prospection. Discovery of L-shaped subsurface anomaly, 2021–23 survey. https://onlinelibrary.wiley.com/doi/full/10.1002/arp.1940
10. Smithsonian Magazine (2024, May). "Scientists Are Investigating a Puzzling Underground 'Anomaly' Near the Giza Pyramids." Coverage of the Western Cemetery GPR/ERT findings. https://www.smithsonianmag.com/smart-news/alongside-egypts-great-pyramid-archaeologists-find-unmarked-underground-structures-180984355/
11. Helal, K. et al. (2025). "Detection of two anomalies behind the Eastern face of the Menkaure Pyramid using a combination of non-destructive testing techniques." NDT & E International, 155, 103331. Archaeology News coverage (November 2025)
12. Lachat, P. et al. (2025). "Investigation of the North Face Corridor in the Great Pyramid of Giza using Electrical Resistivity Tomography." Scientific Reports, 15, 41187. ERT confirmation of ScanPyramids NFC. https://www.nature.com/articles/s41598-025-29081-4
13. Jerusalem Post (2025, March 30). Expert responses to the Malanga/Biondi/Mei Giza SAR claims — Hawass and Abdel-Basir quoted on depth limitations of GPR. https://www.jpost.com/archaeology/archaeology-around-the-world/article-847207
14. Shaaban, F.A. et al. (2009). "Ground-penetrating radar exploration for ancient monuments at the Valley of Mummies — Kilo 6, Bahariya Oasis, Egypt." Journal of Applied Geophysics. 105 gilded mummies, GPR-guided excavation. https://www.sciencedirect.com/science/article/abs/pii/S092698510800164X
15. SPH Engineering / Northrop Grumman (2024). "GPR and Magnetometry Drone Surveys Reveal Unexplored Archaeological Features at Remote Mimbres Site." CulturalSITEs initiative. https://www.sphengineering.com/news/gpr-and-magnetometry-drone-surveys-reveal-unexplored-archaeological-features-at-remote-mimbres-site
16. Ivanova, S. et al. (2020). "Using Ground-Penetrating Radar to Reveal Hidden Archaeology: The Case Study of the Württemberg-Stambol Gate in Belgrade (Serbia)." PMC / Remote Sensing. https://pmc.ncbi.nlm.nih.gov/articles/PMC7037787/
17. Tk'emlúps te Secwépemc Nation (2021). Announcement of 215 GPR-identified suspected burial sites at Kamloops Indian Residential School, May 27, 2021. Wikipedia overview of Canadian residential school gravesites: https://en.wikipedia.org/wiki/Canadian_Indian_residential_school_gravesites
18. APTN News (2023). "How ground-penetrating radar finds unmarked graves." Interview with Dr. Kisha Supernant, University of Alberta Institute of Prairie and Indigenous Archaeology. Includes technical description and community protocol details. https://www.aptnnews.ca/national-news/how-ground-penetrating-radar-is-used-to-find-unmarked-graves-at-residential-schools/
19. CBC News (2025, January 16). "114 'unmarked burial features' detected on former McIntosh residential school property in northwestern Ontario." Wiikwogaming Tiinahtiisiiwin Project Team release; technical lead Aaron Mior quoted. https://www.cbc.ca/news/canada/thunder-bay/mcintosh-indian-residential-school-search-unmarked-burial-features-detected-1.7433302
20. CBC News (2025, January 22). "After 'burial features' detected on McIntosh residential school grounds, researchers help families get closure." https://www.cbc.ca/news/canada/thunder-bay/after-burial-features-detected-on-mcintosh
21. CBC News (2024, April 13). "What comes next in the search for missing residential school children?" Ahousaht First Nation Phase 1 GPR findings. https://www.cbc.ca/news/canada/british-columbia/ground-penetrating-radar-unmarked-graves-residential-school-1.7171154
22. Musqueam Nation / UBC Anthropology (2024). "How Musqueam became a leader in ground penetrating radar." Musqueam-UBC partnership from 2006; community-led protocols; Dr. Andrew Martindale quoted. https://www.musqueam.bc.ca/musqueam-leading-fn-ground-penetrating-radar/
23. Office of the Independent Special Interlocutor (2024, November 3). Upholding Sacred Obligations: Final Report. Federal report on missing children and unmarked burials. Summary and documentation: cedarvia.ca residential schools update
24. International Commission on Missing Persons (ICMP) (2024, March 12). Interim recommendations on missing children and unmarked burials at former residential school sites. Referenced in: cedarvia.ca residential schools source list
25. US Department of the Interior (2023, December). "Final Rule: NAGPRA Systematic Processes for Disposition or Repatriation." Regulations effective January 12, 2024. Federal Register, 88 FR 86188. https://www.federalregister.gov/documents/2023/12/13/2023-27040
26. ProPublica (2025, February). "ProPublica Updates Native American Repatriation Database." 2024 third-highest year for repatriations; 10,300+ ancestors returned; Ohio History Connection holdings. https://www.propublica.org/article/native-american-remains-returned-repatriation-nagpra
27. Bureau of Indian Affairs (2024). NAGPRA compliance status: 2,612 ancestors and 35,826 funerary objects repatriated as of 2024. Official BIA release. https://www.bia.gov/service/nagpra
28. American Bar Association (2024). "The Native American Graves Protection and Repatriation Act: Where Are We Now?" Human Rights Magazine, January 2024. https://www.americanbar.org/groups/crsj/resources/human-rights/2024-january
29. Casini, L. et al. (2023). "A human-AI collaboration workflow for archaeological sites detection." Scientific Reports, 13, 8699. GPR convolutional neural network for Mesopotamian burial detection. https://www.nature.com/articles/s41598-023-36015-5
30. GSSI (Geophysical Survey Systems, Inc.) Technical interview: Peter Leach on GPR for archaeology — depth, resolution, workflow. https://www.geophysical.com/using-ground-penetrating-radar-archaeological-sites
31. Vickers, R.S. et al. (1976). First archaeological GPR survey, Chaco Canyon, New Mexico, 1974–76. Referenced throughout subsequent literature; first European application at Hala Sultan Tekke, Cyprus, 1980: Fischer, P. et al. "Hala Sultan Tekke Revisited." ArcheoSciences (2013). https://www.researchgate.net/publication/261377133
32. Goodman, D. (1994). "Ground penetrating radar simulation in engineering and archaeology." Geophysics, 59(2), 224–232; Goodman, D., Nishimura, Y. & Rogers, J.D. (1995). "GPR time slices in archaeological prospection." Archaeological Prospection, 2, 85–89. Foundational digital time-slice methodology. Commentary: Conyers, L.B. https://gpr-archaeology.com/larry-focus-history-and-goals/
33. Linford, N., Linford, P., Martin, L. & Payne, A. "Stepped frequency GPR survey with a multi-element array antenna: Results from field application on archaeological sites." ArcheoSciences. 3d-Radar GeoScope, 100 MHz–2 GHz, up to 63 channels, <0.1 m line spacing, SFCW architecture. https://journals.openedition.org/archeosciences/1763
34. Linford, N. et al. (2016). "Recent results from a continuous wave stepped frequency GPR system using a new ground-coupled multi-element antenna array." EGU General Assembly / NASA ADS. GeoScope MkIV ground-coupled array, cross-line <0.1 m, archaeological feature detection at landscape scale. https://ui.adsabs.harvard.edu/abs/2016EGUGA..18.4217L/abstract
35. Trinks, I. et al. (multiple, 2013–2022). Multi-channel GPR array advances: 8 cm cross-line spacing, 25 cm post-hole detection, Roman hypocaust brick pillar imaging. Cited in: Nakamura et al. 2024 Archaeological Prospection (source 9 above); Hala Sultan Tekke revisit paper.
36. Qin, T., Bohlen, T. & Allroggen, N. (2023). "Full-waveform inversion of ground-penetrating radar data in frequency-dependent media involving permittivity attenuation." Geophysical Journal International, 232(1), 504–522. FWI resolution ~0.25 m; Ettlinger Line trench detection at Rheinstetten. https://academic.oup.com/gji/article/232/1/504/6670780
37. Feng, D. et al. (2024). "Full-Waveform Inversion of Multifrequency GPR Data Using a Multiscale Approach Based on Deep Learning." IEEE Transactions on Geoscience and Remote Sensing. Deep learning FWI reducing computational overhead for near-real-time field use. https://ieeexplore.ieee.org/document/10488853/
38. Roncoroni, G. et al. (2025). "A realistic 2D multi-offset, multi-frequency synthetic GPR data set as a benchmark for testing new algorithms." Scientific Data (Nature). Multi-offset GPR, full-waveform inversion, pre-stack depth migration benchmark. https://www.nature.com/articles/s41597-024-04300-1
39. Safe2Core Inc. (2024). "The History of Ground Penetrating Radar (GPR)." Covers 1910 patent, 1929 first field use, 1972 Apollo 17 ALSE, 1974–76 Chaco Canyon, 1980s commercial systems. https://safe2core.com/the-history-of-ground-penetrating-radar-gpr/
40. US EPA (2025). "Ground Penetrating Radar (GPR)." Technical overview of frequency/depth/resolution trade-offs, antenna configurations, and environmental limitations. https://www.epa.gov/environmental-geophysics/ground-penetrating-radar-gpr

© 2026 Archaeology Today  ·  All rights reserved  ·  Reproduction with attribution permitted  ·  ISSN 0278-2626 (print) / 1557-9484 (online)