KI positiv nutzen

One third of germanys land area is covered by forests. Beyond their economic importance in the forestry, forests perform vital ecological functions. However, in recent years, rising temperatures and the resulting surge in bark beetle activity have posed significant challenges to these essential ecosystems. Addressing these issues is crucial for preserving our forests and their invaluable contributions to both nature and society.

“The forest is a mirror of society”

After World War II, Germany faced a high demand for building materials. To meet this demand, large-scale monocultures of spruce were planted, as these trees grow quickly and uniformly. While this was economically advantageous, it has led to severe long-term ecological problems.

Since 2018, 500,000 hectares of forest have disappeared in Germany – an area equivalent to more than 700,000 football fields. 80% of this loss is due to the excessive activity of the bark beetle.

The bark beetle

The bark beetle is a naturally occurring insect that becomes active at temperatures above 16 degrees Celsius and thrives from April to September. Under normal circumstances, it aids in the natural rejuvenation of forests by decomposing weak and old trees. However, increasingly drier summers make it difficult for shallow-rooted spruces to access sufficient water and produce resin, their natural defenses against the bark beetle.

Currently, there is no effective method to combat bark beetle-induced spruce die-off. Foresters attempt to identify infested trees visually, a time-consuming and imprecise task. Signs of infestation, like bore dust or reddening crowns, often appear only after the beetles have spread to surrounding trees. In some cases, pesticides are used as a last resort to prevent further tree deaths.

“In the environment, everything is sound”

Bioacoustics, a relatively new scientific field, analyzes environmental relationships through sound. Almost every organism produces sounds for communication, mating, or simply moving. These sounds can reveal intricate natural relationships and enhance our understanding of ecosystems. Applied to the bark beetle problem, bioacoustics offers two key potentials: detecting beetle sounds before visible infestation signs appear and identifying woodpecker activity in winter to locate residual infestations. AI can then analyze and interpret this data, providing valuable insights for managing bark beetle threats.

Detecting Bark Beetle with AI

To identify sounds, audio recordings are converted into spectrograms, which visually represent frequencies over time. AI then applies image recognition principles to these audio data for classification. Developing a Convolutional Neural Network (CNN) for this purpose requires extensive and diverse training data. High-quality data must cover all possible scenarios, including various background noises like wind, rain, cars, running water, rustling leaves, voices, and other animals. This ensures the AI can accurately detect bark beetle activity amidst common environmental sounds.

Amber is an AI-supported monitoring device that records natural sounds throughout the year and detects an infestation of bark beetles before it becomes externally visible.

Depending on the season, Amber uses different natural indicators for successful identification of infested trees. In the summer months, it is deployed in areas with a high risk of bark beetle infestation, such as slopes, south-facing areas, or forest edges that are warmer and drier. It is attached to a reference tree likely to be first infested due to characteristics like crooked growth, smaller crowns or old age.

In winter, when the bark beetles are inactive, it instead measures the feeding behavior of their natural predators, the woodpeckers. Using artificial intelligence, it identifies the specific pecking sounds and calculates the exact location of the source through trilateration. Multiple devices positioned at different locations in the forest record the woodpecker sounds, determining the time of arrival at each device and calculating the differences. Using these timestamps and GPS data, Amber pinpoints the location.

The AI performs calculations locally on the microcontroller. Only positive data, indicating successful identification of either woodpeckers or bark beetles, is forwarded to the forester. This minimizes both transmission costs and data storage requirements. Devices positioned throughout the forest communicate via a LoRa network, which enables extremely efficient and long-range data transmission. This network also bridges any connectivity gaps by allowing each device to act as a hotspot for neighboring devices. Only a master node requires access to the mobile network, where a SIM card is inserted to transmit data to the forester.

At the mid-year mark, Amber undergoes maintenance and inspection. This includes battery replacement, addressing any potential defects or cleaning, and repositioning the device due to changes in measurement needs. Amber features two settings: one activates the rear needle and vibration sensor, while the other utilizes the front microphone for woodpecker localization.

Amber is attached to the trees with a needle in the tree’s bark that also detects the fine vibrations of beetle feeding sounds to a piezosensor and a retried climbing rope that no longer meet human safety standards but are suitable for this purpose. The directionless shape, divided into upper and lower shells, is weatherproof sealed using a press fit over a silicone ring secured by a small rope mechanism. A sliding knot adjusts the climbing rope around the tree, applying force to the small rope, pressing the shells together. It secures or loosens the device without leaving excess rope that could pose a hazard to wildlife.

To protect Amber from damage or theft, integrated barriers have been designed. Firstly, the overall value of the components is low. Additionally, the attachment mechanism is concealed behind the casing, making it less visible and inviting for unauthorized removal. If removal still occurs, a built-in LED starts flashing once the device leaves its location without a Bluetooth connection to the forester’s device.

Impact of Amber

Based on the estimated calculations, each Amber device produces approximately 4.25 kg of CO2-equivalents. For comparison, a hectare of forest sequesters 12 tons of CO2 annually. To remain CO2-neutral, theoretically, up to 2800 devices could be deployed per hectare. However, realistically, only one device is often needed per 1.5-kilometer radius of forest of forest, determined by the spread of the woodpecker’s hacking sound.

Therefore, it can be confidently stated that Amber makes a positive contribution to forest conservation efforts.

The Amber project resulted from the previously developed design method Melior. This approach helps designers navigate in the field of new technologies and identify meaningful applications. Melior is a design method for exploring the potential of emerging technologies like AI. It encourages incorporating these new solutions into design practices. Positioned before the actual creation of a solution, this method helps identify both the potential and limitations of technology within the design process. The developed guidelines can be read here: