An aircraft fires laser pulses at the ground and times how long each one takes to return. Distance is the travel time multiplied by the speed of light and halved. Millions of those measurements make a point cloud, and classifying each return as ground or vegetation and keeping only the ground gives a bare-earth terrain model.

Does lidar see through trees?

No. Some laser pulses pass through gaps in the canopy and reach the ground, and the returns that hit vegetation are filtered out afterwards. The trees are removed by classification, not penetrated. Where the canopy is dense enough that few pulses reach the soil, the resulting terrain model is interpolated and small features are lost.

Can lidar date what it finds?

No. Lidar records shape and height. Dating requires excavation, association with dated material, or comparison with securely dated features of the same form. A lidar survey that reports the age of what it found is reporting somebody else's interpretation.

Can lidar find buried structures?

Only where the buried structure has an effect on the ground surface, such as a mound, a hollow or a differential in how a field has been ploughed. A structure that is genuinely flush with the surrounding ground produces no relief and cannot be detected by any terrain model at any resolution.

What lidar data can I use for free?

In England, the Environment Agency LIDAR Composite Digital Terrain Model is published under the Open Government Licence v3, which permits commercial use with attribution. That is the dataset behind every one of our own surveys.

How much of the corpus? (What is lidar, and what can it not see?)

For England, most of it. The Environment Agency composite covers the large majority of the country at 1 metre. Coverage elsewhere varies enormously, and a jungle survey usually covers a specific project area rather than a region.

What was recovered? (What is lidar, and what can it not see?)

Ground height, and nothing else. Every interpretation on top of that is added by a person or an algorithm afterwards.

What did they say about what they could not see? (What is lidar, and what can it not see?)

This is where most published lidar surveys are weakest. Ground return density under canopy, and the feature classes the detector is blind to, are rarely reported alongside the discoveries.

Did anybody check it independently? (What is lidar, and what can it not see?)

Open data makes it possible. Very few surveys are actually rerun by a second team, and a rerun that used the same settings would not be much of a check anyway.

Method · Lidar

What is lidar, and what can it not see?

Lidar measures ground height by timing laser pulses from an aircraft. Filter out the returns that hit vegetation and you get a model of the bare earth under the trees. It sees shape, not age, not material and not anything flush with the surrounding ground. It also cannot tell you what a shape is.

Environment Agency 1 m composite, Open Government Licence v3

An aircraft flies a grid. It fires laser pulses at the ground, hundreds of thousands a second, and times how long each one takes to come back. Distance is time multiplied by the speed of light, halved. Do that densely enough and you have a cloud of points, each with a height.

The archaeology happens in the next step, and it is a filtering step rather than a measuring one. Some pulses hit leaves. Some hit branches. Some make it through a gap in the canopy and hit soil. Classify each return as ground or not-ground, throw away the not-ground, and interpolate a surface through what remains. That surface is a digital terrain model, and it is the picture people mean when they say lidar found a city under the jungle.

Nothing was seen through anything. The trees were removed arithmetically, from a small number of pulses that happened to reach the floor between them.

What that means for what survives

Ground return density is the whole game under canopy. Dense broadleaf in leaf-on conditions can drop the fraction of pulses reaching the soil to a few per cent, and a terrain model interpolated across those gaps is smoother than the real ground. Small features vanish into the interpolation before any detection algorithm gets near them.

That is why survey flights over forest are flown in winter where the trees are deciduous, and why a survey over evergreen canopy is a harder result than the same survey over oak.

The four things it cannot do

It cannot date anything. A bank is a bank. A medieval field boundary and a Bronze Age enclosure are the same kind of shape to a height model, and the only thing that separates them is context somebody supplies afterwards.
It cannot see anything flush with the ground. A buried wall with no surface expression, a filled ditch that has been ploughed level, a floor surface: no relief, no signal. Ever, at any resolution.
It cannot tell you what a shape is. This is the one that produces most of the bad coverage. A detector returns positions and dimensions. Whether the thing at that position is a barrow, a spoil heap, a slurry pit or a silage clamp is a separate question that lidar does not answer.
It cannot see a depression, if you built it to find mounds. Not a limitation of lidar, a limitation of most detectors people build on top of it, and ours had it too. That has its own page.

Refuted

On our first survey over 6 square kilometres of Dorset, the three highest-ranked candidates the detector produced were all modern. A linear embankment that is a road cutting, and two sets of farm buildings. The algorithm was working correctly. It finds raised shapes of roughly the right size, and a modern farmyard is a raised shape of roughly the right size.

Run 001, Dorset downland block A, 5 August 2026. Top three candidates by rank, triaged visually: a road cutting, a farmyard, and farm buildings.

What we use, and why the licence matters

Everything on this site that comes off our own runs uses the Environment Agency LIDAR Composite Digital Terrain Model at 1 metre, published under the Open Government Licence v3. It covers most of England, it is free, and the licence permits commercial reuse with attribution.

That last point does more work than it looks like it does. It means every plate in book 1 can be printed and sold, and it means anyone can rerun what we did over the same ground and get a different answer from ours if we have made a mistake. A survey nobody can repeat is not a result, it is an announcement.

The four questions, applied

The same four we put to every result on this site, turned on this method.

How much of the corpus?: For England, most of it. The Environment Agency composite covers the large majority of the country at 1 metre. Coverage elsewhere varies enormously, and a jungle survey usually covers a specific project area rather than a region.
What was recovered?: Ground height, and nothing else. Every interpretation on top of that is added by a person or an algorithm afterwards.
What did they say about what they could not see?: This is where most published lidar surveys are weakest. Ground return density under canopy, and the feature classes the detector is blind to, are rarely reported alongside the discoveries.
Did anybody check it independently?: Open data makes it possible. Very few surveys are actually rerun by a second team, and a rerun that used the same settings would not be much of a check anyway.

Sources

LIDAR Composite Digital Terrain Model, 1 metre. Environment Agency, Open Government Licence v3, 2026.

Where this is written up in full

Lost Under the Canopy

Six surveys, 907 candidates, one confirmed. Then the same four questions turned on the most famous survey results in the world.

All four are written and none is on sale. Advance readers can read them first, in exchange for an honest review.