Applied AI & AI in business /

Minimising the spread of bark beetle infestation in forests all around Europe

May 20/4 min read
  • Lara Agneter, Irem Eda Dönmez, Kaja Martyna Kowalczyk, Kilian Maximilian SorgWinners of the AI for Business Challenge

Bark beetles are insects that are just millimeters long, spreading across forests - particularly in Europe and North America. By infesting a tree the bark beetle blocks the circulation of sap by attacking the bark of spruce trees and boring holes into the tree to lay its eggs, from which emerge larvae that attack the bark.

This results in the tree’s death in less than four weeks.

These small creatures can cause dramatic, irreversible alterations both in natural and urban forest environments. Although bark beetle infestations are a regular force of natural change in forested ecosystems, several of the current outbreaks are the largest and most severe in recorded history due to weakened woodlands caused by rising heat and drought. 

02/ What can be done to minimise bark beetle infestations

In Germany and France especially, the bark beetles are currently threatening many forests. In the US the beetle epidemic already had some severe consequences like wildlife and environmental threats. This can lead to further long-term problems like threatening water resources, which results in the absence of sheltering trees. In the US it also led to grizzly bears moving closer to villages as well as becoming an endangered species.

As soon as one tree is infected, it should be recognized, cut and brought away from the forest as soon as possible to minimize the spread of the bark beetles. In the examples of Germany, France and the US, this spread was not stopped early enough and is now a major environmental and societal concern. 

We have built an AI model with the purpose of helping to detect infected trees faster and thereby, able to stop the spread of bark beetles more efficiently.

Bark beetles are currently threatening many forests in Europe and North America

03/ Data collection

The AI model analyses pictures of tree trunks and scans for potential bark beetle infestation. The data collected for the MVP was pictures mostly taken from the Black Forest in South Germany. 

For our prototype, we would set up 360° cameras on moving objects (i.e. forest vehicles, drones, etc.) of forest workers to collect valuable data of the forest stand as well as use existing cameras that are already set up in many mountain areas. 

In the future, we see high potential in the use of drones. Instead of analyzing the trunks of the tree, the drone would analyze the crones to identify individual infected trees, as the crown’s color and texture says a lot about if a tree is infected or not.

04/ How the AI model works

The project aims to support forest workers in discovering infected trees as early as possible to better contain the rapid propagation of bark beetles. Earlier identification of infected trees enables forest workers to cut the contaminated trees and bring them outside of the forest to prevent the bark beetles from infecting other trees. Time is of the essence due to the rapid growth of bark beetles and their ability to kill an infected tree within only four weeks. 

The AI identifies infected trees and locates their exact position based on the coordinates associated with the metadata of the recordings. The trees are analyzed and categorized as (A) infected or (B) healthy. 

Infected trees can be identified by various transformations in their trunk and crown area such as:

Trunk:

  1. Concentrations of brown bore dust at the basal area of the trunk
  2. Bark falling off
  3. Holes in trunk area

Crown:

  1. Discoloration starting at the bottom of the crowns and moves upwards
  2. Needles dropping from the crown when green

These changes enable our AI to cluster trees into two categories. As the recordings capture the coordinates the forest workers can easily identify the exact location.

Infected trees can be identified by various transformations in their trunk and crown area

05/ The inspiration behind the cause

Within the project team, we have a personal connection to forest workers in the Black Forest (Forest Area in South Germany) and through priorly conducted interviews, the widespread issue of bark beetles became evident. 

The foresters highlighted that it is a huge struggle to identify the harmed trees as fast as possible and unfortunately in most cases, the identification of a damaged tree happens too late in the infection process and thus, they are forced to cut a lot of trees down to prevent the bark beetles from spreading to healthy trees nearby.

Further on, the forest workers also talked about the decreasing income for the timber industry, as reforesting is costly. One hectare of planting new trees costs between €5,000 and €10,000. While town/municipality-owned forests have the money to do this, private forest owners will not be able to afford this, which poses a major problem for individuals or families who own forests and can be even threatening their livelihood.

06/ Conclusion

The AI tool helps forest workers detect the infected trees earlier and faster before it spreads out to other trees. Measuring the change of infected trees in a forest allows forest workers to track the growth and decline of bark beetle infections throughout time. Later on, the AI tool could be improved with more pictures added, as well as with improved technological capabilities to scan the forests faster and more accurately

Lara, Irem, Kaja and Kilian of the winning team

  • Lara Agneter, Irem Eda Dönmez, Kaja Martyna Kowalczyk, Kilian Maximilian Sorg

    Winners of the AI for Business Challenge

    Lara, Irem, Kaja and Kilian are studying Innovation & Sustainability at Copenhagen Business School by day and building AI models by night. As the winners of the AI for Business Challenge taking place in April 2021, the group built an AI model to help minimize the spread of bark beetle infestation in Northern Europe.

07/ More on Business & Applied AI