Skip to content

Service B · 1 · Archaeological prospection

LiDAR archaeology. Prospection before the spade.

Laser scanning measures the ground even beneath dense forest. We turn it into relief visualisations, screen large areas with AI and review every candidate before anyone goes into the field.

Abb. 01LiDAR terrain model, 1.2× vertical exaggeration · data: LGL BW, dl-de/by-2.0, modified

01What shows up

Forest in the aerial image. History in the relief.

A detail from our hero lens: beneath the canopy lie landforms no aerial image shows. The labels deliberately reveal no location.

Terrain model of a woodland with marked hollows and a round mound
  1. 01Funnel-shaped pits in the forestDozens of hollows, invisible in the aerial image. Interpretation open: sinkholes or material pits – only review will tell.
  2. 02Low round moundInvisible in the aerial image, clear in the relief. Interpretation and location are deliberately left open.

LiDAR terrain model · Data: LGL BW, dl-de/by-2.0, modified

02Typical structures

What relief maps make visible.

  • 01

    Burial mounds

    Round mounds, often in groups – preserved for millennia under forest.

  • 02

    Hollow ways

    Sunken old roads, often as bundles on slopes.

  • 03

    Ridge and furrow

    Parallel strips of former fields beneath today’s forest.

  • 04

    Terraces

    Steps of former vineyards or fields on slopes.

  • 05

    Charcoal kiln platforms

    Circular platforms left by charcoal production.

  • 06

    Banks and ditches

    Linear structures of fortifications, boundaries or earthworks.

03Method

Several views, one reasoned list.

No relief visualisation shows every structure. For each area we compute the same verified set of six views, let the AI screen it tile by tile and review every candidate blind – without knowing location or expectation.

  • Rejections are reasoned: forest road, skid trail, timber yard, recent use
  • Cross-checked against historic maps and aerial images
  • Open candidates with a survey recommendation – never called a “find”

04Responsibility

Sites deserve protection.

Finding candidates comes with responsibility. These rules apply to every project.

  1. 01No coordinates published – not on this site, not in talks
  2. 02Results go to the client and the responsible heritage authority
  3. 03Field surveys only with permission and by the responsible specialists
  4. 04Internal metrics are labelled as internal and not externally validated

05Exchange

Method in the open.

We presented our workflow at the CAA Joint Chapter Meeting in Münster in September 2026 and opened it up for discussion.

  • Blog

    Context Before Pixels

    The talk in writing (German): how a multi-agent workflow turns open LiDAR data into a verified candidate list.

  • Blog

    AI judgements in archaeology

    How a typed AI model reasons about classifications – three practical examples (German).

06Fragen

Frequently asked.

Does LiDAR replace field survey?

No. LiDAR and AI narrow down the area and give reasons where a survey is worthwhile. Whether anything is there is for specialists in the field to establish.

What resolution do we need?

Small structures such as charcoal platforms or ridge and furrow need a terrain model of 1 m resolution or finer. Many regions publish such data openly; we check before the project starts.

Do you work with heritage authorities?

Our results are intended for heritage management and go to the responsible authority. We publish no coordinates and recommend field surveys only with permission.

How reliable is the AI screening?

We measure it per project: with blind review, reasoned rejections and – where available – known monuments as a control. We do not promise general hit rates.

Which area do you want to pre-screen?

Tell us the area and the research question. We will tell you what data exists and what an analysis can deliver.