Dubalang: When Trees Have Ears

Forest protection has long been closely associated with patrols. Rangers and local communities walk through undergrowth, follow rivers, read tracks, and rely on their eyes to identify signs of disturbance. This approach remains important, but it has its limitations. Forests are vast and dynamic: wildlife moves, people come and go, vehicles can be heard from a distance, and activities that threaten an area may take place far beyond the reach of patrol teams.

To identify more quickly what is happening inside forests, KKI Warsi and Rainforest Connection (RFCx) developed Guardian, an audio-based monitoring system that combines acoustic technology and artificial intelligence. Devices are installed in trees to capture surrounding sounds, then transmit information through communication networks so that it can be analysed and received by rangers. As the technology has developed, it has been used to recognise a range of sounds, including chainsaws, vehicles, gunshots and wildlife, enabling rangers to receive information in real time and determine the appropriate follow-up action.

Experience with Guardian has provided an important lesson: technology does not replace patrols, but expands people’s ability to monitor forests. Even when people are not present at a particular location, the devices can continue listening to what is happening nearby. Alerts received by rangers are then verified to establish the type of sound and determine whether an on-site inspection is necessary. In several nagari forests in West Sumatra, the system has even helped patrol teams discover illegal activities and collect field evidence in the form of audio recordings, photographs, and geographic coordinates.

These lessons were then applied in a different context. While Guardian is widely used to support the protection of community-managed forests, KKI Warsi developed Dubalang for monitoring across a broader and more complex landscape, particularly the Berbak-Sembilang Landscape and Berbak and Sembilang National Park. The name Dubalang has close ties to Sumatran traditions, referring to a guard or protector. In the context of this technology, the name symbolises the idea of having a ‘guardian’ that keeps listening when people are not inside the forest.

The choice of this landscape was not accidental. Berbak-Sembilang is an important natural landscape containing wetland ecosystems, including peat swamps and coastal forests, whose ecological functions extend far beyond the boundaries of protected areas. Berbak and Sembilang National Park is also an important habitat for the Sumatran tiger (Panthera tigris sumatrae), an endangered endemic species included among the 25 national priority species.

Yet Berbak-Sembilang’s importance is not solely due to its tigers. The landscape forms part of a wider hydrological system. Buffer areas such as Orang Kayo Hitam Forest Park (Tahura) play an important role in regulating water flows across peatland areas, while also storing water and carbon and providing habitat for a wide range of flora and fauna. Consequently, when forests and peatland ecosystems are disturbed, the impacts do not remain within the affected area. Changes to water flows, fires, habitat loss and ecosystem degradation can affect the wider landscape, including the lives of communities downstream.

In these circumstances, the ability to identify disturbances as early as possible is crucial. This is not because patrols are no longer needed, but because the size and character of the area make it impossible for people to be everywhere at the same time. Technology is needed to reduce the areas left unmonitored and speed up the flow of information when something happens. This is where Dubalang comes in.

Dubalang builds on the lessons learned from Guardian by combining acoustic sensors, the Internet of Things (IoT), artificial intelligence (AI), telecommunications networks, and solar panels in a single monitoring system. Devices are installed high in trees so that they can capture sounds from the surrounding environment. Solar panels provide the power, while communication networks allow the information captured by the devices to be sent to designated recipients.

The way it works is relatively straightforward. Dubalang listens to and captures surrounding sounds through acoustic devices. When a particular sound is detected, information is sent over a telecommunications network to a phone number that has been pre-set as a recipient. The recipient can listen to the sound in real time and receive information about what the device has picked up.

This is where AI adds value. Forests generate a vast amount of sound, and not every sound has the same significance for those managing the area. AI helps identify sound patterns and classify captured audio, such as chainsaws, vehicles, engines, and animal calls. This allows people to focus their attention on sounds considered important without having to listen manually to every recording.

However, the results of sound recognition do not automatically constitute a conclusion. For example, if Dubalang captures a sound identified as a possible chainsaw, the alert received by rangers still needs to be verified. Rangers can listen to the sound again, assess the information, and then decide whether the location needs to be checked in person.

At this point, Dubalang’s role in the forest protection system becomes clear. It is not an ‘automated forest police officer’, let alone a replacement for people. Dubalang is an early-warning system that shortens the gap between an incident in the forest and the people authorised to respond to it.

Even a small improvement in response time can make a significant difference. An activity that has only just begun is very different from an area that has already been damaged. When information about the sound of machinery is received earlier, rangers have a better chance of deciding on the right course of action before a disturbance develops into more extensive damage.

Dubalang therefore makes patrols more information-led. Patrols do not always have to follow routine patterns or wait until signs of disturbance are discovered in the field. Information from the devices can help determine which locations need to be inspected and when inspections should take place. In this way, patrol personnel, time and costs can be used more effectively.

This approach can also complement other monitoring systems already developed in Berbak-Sembilang. Camera traps provide information about the presence and movements of wildlife, while SMART Patrol helps record rangers’ activities and field observations. Dubalang adds another dimension by continuously capturing sounds from specific locations.

Each of the three serves a different purpose, but they can complement one another. Cameras help reveal what people cannot see; Dubalang hears what is difficult to hear from a distance; and patrol data connect technological information with actual conditions on the ground. When these data are managed together, the monitoring system becomes more than a tool for detecting disturbances: it also becomes a source of information for understanding changes across the landscape.

One particularly interesting aspect of Dubalang’s development is the potential to use forest sounds for purposes far beyond detecting chainsaws. Forests have a complex soundscape, comprising birds, mammals, insects, frogs, flowing water, rain, wind and human activity. When these sounds are recorded continuously, there is an opportunity to use them to build knowledge about ecosystem conditions. AI can help process large volumes of data that would be difficult to analyse manually, allowing acoustic technology to evolve from a tool for detecting threats into another way of understanding forest life.

Yet technological sophistication does not mean that every problem is solved. Dubalang still requires telecommunications coverage, sufficient power, well-maintained equipment and recipients who are ready to respond. Most importantly, there must be a clear follow-up system. An alert will mean little if no one checks and verifies it or makes a decision after receiving the information.

Technology must therefore remain part of a broader system. Behind every device installed in a tree, people, institutions, procedures, patrols, coordination and the authority to take action are still needed. Technology speeds up the flow of information, but people must still determine what that information means and what should be done.

This is especially important in the Berbak-Sembilang Landscape, where management involves many stakeholders. The national park, local government, surrounding communities, managers of buffer areas, law-enforcement agencies, researchers and conservation organisations each have their own roles. Developing wildlife-monitoring strategies, including monitoring of Sumatran tigers, requires collective effort so that the data gathered do not remain merely information, but can be used to strengthen area management.

In this context, Dubalang can serve as a link between technology and field response. When a sound is detected, the information is received by designated parties, verified and then used as the basis for deciding what to do next. The shorter the interval between an incident and the response, the greater the opportunity to prevent a disturbance from escalating into more extensive damage.

This is where the idea of ‘trees having ears’ becomes more than a metaphor. A tree, once seen simply as an object to be protected, can now host a device that listens to its surroundings. When a sound indicating a potential threat is heard, the information can be transmitted out of the forest to someone who is able to respond.

Technology, then, is not the end in itself. It is a new way to strengthen an enduring responsibility: protecting forests so that they remain a home for wildlife, a safeguard for water systems and a source of life for communities.