Discovering cultural heritage using laser data

The availability of new data allows us to advance our position in forest management. Now, a Mistra Digital Forest project aims to develop methods for detecting cultural remains using laser data, which would simplify and improve the work of protecting these historic sites. 

There are strict requirements for forest owners to protect cultural heritage sites during felling. At the same time, many of these are difficult to avoid making it quite a challenging responsibility to live up to. Now we want to see if we can use laser data to detect cultural remains more easily, says Henrik Persson, senior researcher in digitalisation at Skogforsk.

Recent studies show that it is possible to detect charcoal burning sites, trapping pits and other large cultural remains on the ground using Lantmäteriet's national laser data. However, so far it has not been possible to detect remains such as house foundations and graves with the available national data. 

- At the same time, it is precisely these kinds of remains that the industry wants digital support to detect, says Henrik. Since the national laser data is currently insufficient, as a first step we want to see if we can identify these types of cultural remains with the help of more high-resolution data. 

Aiming to produce a map of cultural remains in all of Sweden's forests  

A 2200 hectare test area has been laser scanned in three wavelengths, with high point density. By classifying the points - that is describing what they represent and grouping them - the detailed data can be used to train AI models. Field inventories from the same area are used as reference material. 

One challenge is that the remains are so geographically dispersed. In the test area there were about a hundred similar cultural remains, and more extensive data is required to train an AI model. The solution to this problem could be found in the new National Forest Data Lab hosted by Skogforsk, which collects data from the entire forestry sector. There, high-resolution data from more forest owners could give the AI model a larger database to train on.  

- We are just starting out, but our aim is to produce a map of cultural remains in all of Sweden's forests. It could provide an additional layer in planning tools for forest management, and the harvester could erect a kind of digital fence around sensitive areas. This would be a clear improvement as compared to today when we manually label remains for example, often based on inadequate data, says Henrik.