The Archaeology We Do Not Dig: Are We Entering a Post-Dig Era?
In the Amazon, airborne lasers are exposing the scale of ancient engineered landscapes; beneath a Danish football pitch, radar has mapped a medieval convent without opening a trench. Archaeology can now locate, map and sometimes interpret entire landscapes while leaving them physically untouched. But seeing buried archaeology and understanding it remain rather different achievements.
In southwestern Amazonia, the forest canopy is no longer the obstacle it once was. In July 2026, Martti Pärssinen and an international team reported the results of an airborne LiDAR programme designed to estimate the distribution of precolonial earthworks across a cultural landscape far larger than archaeologists could realistically inspect on foot. The survey involved 4,430 kilometres of LiDAR flight lines, arranged as long transects through Acre and neighbouring parts of the Brazilian Amazon. More than 400 well-preserved monumental sites were documented in the surveyed sample. From this, combined with existing archaeological information, the researchers estimated that the wider cultural area could contain more than 20,000 earthworks.
That distinction matters. Twenty thousand structures did not suddenly appear as twenty thousand neat outlines on a computer screen. LiDAR sampled a fraction of a territory estimated at roughly 183,000 square kilometres; the larger number is an extrapolation from the distribution observed within that sample. The researchers also proposed a population of approximately 1.25–3 million people for the region around AD 100–300. Those are archaeological models built from impressive new evidence, not direct counts of either monuments or inhabitants.
Yet the change in scale is real. Earlier archaeological knowledge of the geometric earthworks of southwestern Amazonia was heavily biased towards deforested areas, precisely because that was where large ditches, enclosures and roads could most easily be recognized in satellite imagery or from the ground. Canopy-penetrating laser scanning alters the terms of the problem. Rather than waiting for clearance to make archaeology visible, researchers can construct detailed digital terrain models from the small proportion of laser pulses that reach the ground between leaves and branches. The forest remains; the topography underneath becomes legible.
The obvious temptation is to declare this a new kind of archaeology in which excavation is becoming optional.
That would be premature.
What the instrument actually sees
LiDAR is often described as though it sees through vegetation. Strictly speaking, it does nothing of the sort. An airborne scanner sends large numbers of laser pulses towards the surface and measures their returns. Some strike the canopy, some lower vegetation, and some reach the ground. Processing the resulting point cloud allows researchers to filter much of the vegetation and model the underlying terrain.
What appears is therefore topography: banks, platforms, ditches, causeways, terraces, mounds and other modifications that retain sufficient relief to be distinguished from their surroundings. LiDAR can be spectacularly effective where archaeological architecture consists of earth rather than stone, and especially where vegetation has made conventional mapping difficult. In the Bolivian Amazon, for example, Heiko Prümers and colleagues used airborne LiDAR to map Casarabe-period settlements, revealing monumental platforms, pyramidal structures, canals, reservoirs and straight causeways within a hierarchically organised settlement system. The survey documented two particularly large centres of 147 and 315 hectares.
What LiDAR did not establish by itself was when those structures were built, whether different elements were contemporary, what people ate there, what objects they used, or precisely how individual buildings functioned. The Casarabe interpretation rests partly on previous excavations, radiocarbon dating, archaeobotany and zooarchaeology. Excavated evidence showed that monumental sites were inhabited rather than simply visited ceremonially, and that their inhabitants cultivated crops including maize while also hunting and fishing. The aerial map became archaeologically intelligible because other forms of evidence already existed.
The same principle applies to the new southwestern Amazonian survey. LiDAR radically improves the sample from which regional patterns can be reconstructed. It can expose the geometry of a landscape and make possible quantitative arguments that previously rested on a badly distorted distribution of known sites. But a ditch visible in a terrain model has no date attached to it.
The chronology matters here. If two neighbouring enclosures were built three centuries apart, a map that displays them simultaneously may imply a settlement system that never actually existed in that form. Remote sensing is exceptionally good at collapsing space onto a screen. Archaeology must then put time back into it.
A monastery beneath the football pitch
The problem looks different in Roskilde, Denmark.
Historical sources already established that the medieval convent of St Clare had occupied the area now partly covered by a sports field at Byparken. The house dated from the thirteenth century and was dissolved following the Reformation. Large-scale dismantling subsequently supplied bricks for other buildings, and the extent to which its foundations survived was uncertain. When proposed construction of an underground car park made that question immediately practical, archaeologists turned first to ground-penetrating radar.
The survey itself was conducted in January 2025 by specialists from the Norwegian Institute for Cultural Heritage Research and the National Museum of Denmark in collaboration with ROMU. The results, announced in August 2026, were unusually clear. At approximately one metre below the present surface, the radar data trace the monastic ranges around a cloister, room divisions, structural supports and a church in the southern wing. The northern and eastern ranges appear especially well preserved.
This is not simply another form of aerial archaeology. GPR sends electromagnetic pulses into the ground and records reflections produced where materials with different electromagnetic properties meet. Closely spaced measurements can be processed into horizontal slices representing different depths. Under suitable conditions, walls, floors, voids, robbed foundations and other structures can therefore be mapped three-dimensionally.
The qualification “under suitable conditions” deserves emphasis. Soil conductivity, clay content, moisture, surface conditions, depth and the contrast between archaeological deposits and their surrounding matrix all affect the quality of the result. At Falerii Novi, the Roman town north of Rome that became one of the most ambitious demonstrations of large-area archaeological GPR, even rainfall affected penetration because increased soil conductivity reduced radar performance.
Roskilde produced exceptionally legible data. Not every site does.
And even there, the radar image does not reveal whether a particular masonry line belongs to the first construction phase or a later rebuilding. It cannot provide a coin from a foundation trench, carbonised seeds from a floor deposit or mortar suitable for laboratory analysis. Nor can it determine whether two adjacent rooms were used at exactly the same moment merely because both appear clearly in a depth slice.
What it can do is something excavation is poorly suited to doing quickly: show the organisation and preservation of virtually an entire architectural complex before decisions are made about where, whether or how extensively to dig.
That is a profound change in archaeological practice.
From finding sites to choosing questions
The transformation is most obvious when several techniques are combined.
Magnetometers measure minute variations in the local magnetic field. Filled ditches, pits, hearths, kilns and some forms of masonry may produce detectable anomalies. Magnetometry can cover large areas rapidly, but what it detects depends on the magnetic contrast between the archaeological feature and the surrounding geology.
Electrical resistance survey measures how easily a small electrical current passes through the ground. Moist ditch fills commonly conduct electricity more readily than dry stone foundations; seasonal soil moisture can therefore strongly affect results. GPR works according to different physical principles and may resolve depth as well as plan. No single technique is universally superior.
Falerii Novi demonstrates why this matters. Earlier magnetometry had already produced an impressive plan of the 30.5-hectare walled Roman town. High-resolution GPR subsequently revealed information at different depths and identified architectural details, floors and previously unrecorded buildings. Yet comparison between the two datasets showed that some features visible magnetically were weak or absent in the radar results. The researchers explicitly concluded that neither technique produced a complete archaeological picture.
The future, then, is probably not one all-seeing instrument but data fusion.
Satellite imagery can identify crop, soil, moisture and spectral differences across landscapes that would be prohibitively expensive to survey conventionally. Synthetic-aperture radar is being tested for detecting anthropogenic structures even in heavily forested environments. Drones can acquire centimetre-scale photography, multispectral imagery or LiDAR across targeted areas. GIS allows all of these observations to be compared with topography, hydrology, historical maps, previous finds and excavation results.
The volume of information has created another problem: archaeologists may now be capable of collecting data faster than specialists can inspect it.
Machine-learning systems are therefore increasingly being trained to flag probable archaeological features in LiDAR and satellite datasets. Experiments have already targeted burial mounds, charcoal kilns, settlements and other repetitive feature classes. A useful model is not one that replaces archaeological interpretation, however, but one that reduces millions of pixels or terrain cells to a manageable set of candidates for expert inspection and field verification. Training data embody existing archaeological knowledge, including its biases; unusual sites are by definition more difficult for systems trained on familiar ones to recognise. False positives are not a software inconvenience but an archaeological problem if they begin to shape maps of past settlement.
The same caution applies in reverse. An absence of detected features is not necessarily archaeological absence. It may describe the limitations of the sensor, vegetation, geology, soil conditions, survey resolution or classification procedure.
Archaeology without artefacts?
Some of the newest techniques move still further away from the familiar image of excavation.
Ancient DNA can now be recovered directly from archaeological sediments. In 2017, Viviane Slon and colleagues demonstrated that Pleistocene cave sediments could preserve mitochondrial DNA from Neanderthals and Denisovans even in layers where no hominin bones had been found. Subsequent work has retrieved increasingly detailed genetic information from sediments. At El Mirón Cave in northern Spain, sedimentary ancient DNA recently identified 28 animal taxa, including humans, with 15 taxa absent from the previously identified faunal assemblage; human mitochondrial sequences were also recovered from Solutrean deposits.
It would nevertheless be misleading to call this entirely non-invasive archaeology. Sediment has to be sampled. More importantly, DNA is valuable archaeologically only if the provenance of the sample is understood. A molecular result detached from reliable stratigraphy can become remarkably difficult to interpret. Sediments move; water percolates; animals burrow; microscopic material may be displaced. Molecular sensitivity does not abolish taphonomy.
This points to an important distinction. “Archaeology without excavation” can mean at least three different things: genuinely remote observation; non-destructive geophysical survey on the ground; or minimally invasive sampling that removes tiny amounts of material rather than opening large trenches. They should not be treated as methodological equivalents.
What unites them is not an absence of physical evidence but a changing relationship with it.
Excavation after the map
For much of archaeology's history, discovering what lay beneath the surface required removing what lay above it. Excavation therefore combined detection and investigation in a single operation. That is increasingly unnecessary.
At many sites today, archaeologists can first map the larger landscape, recognise probable structures, identify preservation conditions and formulate questions. Trenches can then be placed not simply where archaeology might exist but where excavation can answer a particular chronological, functional or environmental problem.
This reverses an old practical hierarchy. Excavation becomes the targeted test rather than automatically the principal means of seeing the site.
There is also an ethical advantage. Archaeological excavation is destructive. However meticulous the recording, excavated stratigraphy cannot be reconstructed physically once removed. If geophysical survey can answer a question while leaving deposits intact, there is a strong argument for doing so, particularly at vulnerable or protected sites. Large-scale prospection is already used in cultural-resource management to assess buried archaeology before development and to direct any necessary invasive investigation.
But the strongest argument against a literal “post-dig era” comes from the remote-sensing successes themselves.
Falerii Novi was mapped with sufficient detail to identify streets, public buildings and architectural sequences that had never been excavated. Yet subsequent research at the site returned to excavation. This was not an admission that geophysics had failed. It was the logical next stage. Once the town plan was known, excavation could address questions the map could not settle: chronology, construction sequences, occupation deposits, artefact assemblages and the social histories of particular spaces. Recent field seasons have therefore combined the extraordinary geophysical map with targeted trenches and material analysis.
The Amazonian case leads to the same conclusion on a far larger scale. LiDAR may tell us that settlement and landscape engineering were vastly more extensive than a forest-floor survey could reasonably demonstrate. It may reveal roads linking sites and suggest hierarchies among monumental centres. But population estimates still require assumptions about how many sites were occupied simultaneously, how settlements functioned and what demographic densities are appropriate. Claims about agriculture require plant remains, phytoliths, pollen, starches or other environmental evidence. Claims about chronology require dates. Claims about social organisation ultimately depend on more than geometry.
Remote sensing changes where excavation occurs and what excavation is expected to achieve. It does not remove the need for it.
Perhaps “post-dig” is therefore the wrong description of what is happening. Archaeology is moving towards a pre-dig era in a more literal sense: increasingly sophisticated investigation now precedes the trench.
At Roskilde, the radar map can guide decisions about what should be preserved and which small areas might need excavation. In Amazonia, kilometres of laser data can identify the few locations where ground investigation would most effectively test a regional model. At Roman towns, magnetometry and GPR can place individual trenches within an urban plan already visible at city scale. Sedimentary DNA can tell archaeologists which layers may contain molecular evidence even where bones are absent.
The result is not archaeology without digging. It is archaeology in which digging no longer has to carry the entire burden of discovery.
For some questions, the archaeological site can now remain underground.
For the questions that depend on sequence, date, material culture, diet, economy, human remains or the formation of deposits, the trench remains difficult to replace. The important methodological shift is that archaeologists are becoming much better at knowing why they are opening it.
Selected Sources and Further Reading
Pärssinen, M., Kalliola, R., Ranzi, A., et al. 2026. “Over 20,000 precolonial earthworks in the Southwest Amazonia.” Nature. Published 29 July 2026.
DOI: https://doi.org/10.1038/s41586-026-10835-7
Norwegian Institute for Cultural Heritage Research (NIKU). 2026. “Thought the monastery was lost: Found the entire complex with ground-penetrating radar under a football pitch.” Published 12 August 2026, updated 13 August 2026.
Prümers, H., Jaimes Betancourt, C., Iriarte, J., Robinson, M., Schaich, M. et al. 2022. “Lidar reveals pre-Hispanic low-density urbanism in the Bolivian Amazon.” Nature 606: 325–328.
DOI: https://doi.org/10.1038/s41586-022-04780-4
de Souza, J. G. et al. 2018. “Pre-Columbian earth-builders settled along the entire southern rim of the Amazon.” Nature Communications 9.
DOI: https://doi.org/10.1038/s41467-018-03510-7
Verdonck, L., Launaro, A., Vermeulen, F. & Millett, M. 2020. “Ground-penetrating radar survey at Falerii Novi: a new approach to the study of Roman cities.” Antiquity 94(375): 705–723.
DOI: https://doi.org/10.15184/aqy.2020.82
Slon, V. et al. 2017. “Neandertal and Denisovan DNA from Pleistocene sediments.” Science 356(6338): 605–608.
DOI: https://doi.org/10.1126/science.aam9695
Gelabert, P. et al. 2025. “A sedimentary ancient DNA perspective on human and carnivore persistence through the Late Pleistocene in El Mirón Cave, Spain.” Nature Communications.
https://www.nature.com/articles/s41467-024-55740-7
Casini, L. et al. 2023. “A human–AI collaboration workflow for archaeological sites detection.” Scientific Reports 13.
https://www.nature.com/articles/s41598-023-36015-5
Historic England. “Geophysical Survey Techniques: Magnetometer, Earth Resistance and Ground Penetrating Radar.” Current technical guidance.
Suggested keywords
archaeological remote sensing; LiDAR archaeology; ground-penetrating radar; archaeological geophysics; Amazonian archaeology; precolonial Amazonia; archaeological prospection; sedimentary ancient DNA; satellite archaeology; machine learning in archaeology; Falerii Novi; St Clare’s Convent Roskilde
Editorial summary
The 2026 Amazonian LiDAR survey and the newly published GPR mapping of St Clare’s Convent in Roskilde make this an unusually timely moment to ask not whether excavation is disappearing, but whether archaeology is entering an era in which the decision to excavate comes only after much of the site has already been mapped, modelled and interrogated from above or from the surface.