Ultra-Early Wildfire Detection at Waitangi: New Zealand’s First Connected Smart Forest

The collaboration between the Far North District Council (FNDC), Spark New Zealand, Adroit Environmental Intelligence, and Germany-based Dryad Networks marks a major milestone for climate resilience in Aotearoa.

Philippe Boulanger, Business Lead at Adroit, says the project reflects a shared commitment to early intervention. “This deployment demonstrates the power of partnerships, bringing together technology providers, local government, conservation groups, and recreation communities, all working together with the same goal: earlier detection of fires so that first responders can act before fire becomes dangerous or unmanageable.”

The system detects the earliest chemical and thermal signatures of smouldering material — often long before there is visible flame or smoke — and sends alerts across Spark’s national IoT network for rapid response.

For a region increasingly exposed to drought and high winds, the installation represents a step change in environmental protection and a model for how regional authorities can use technology to safeguard land, people, and cultural heritage.

“Spark worked closely with Dryad to make remote connectivity possible,” Boulanger says.

“Extending connectivity into remote terrain is no small task, but it’s essential in enabling this new class of environmental protection.”

Shyam Nathoo, Spark Business Development Manager IoT, says the project showcases a new generation of environmental monitoring technology.

“What makes this solution unique is its ability to detect fire during the smouldering phase, before flames reach the canopy where traditional systems are more likely to identify it. In practical terms, that can reduce detection times from hours to minutes and give responders a much greater opportunity to intervene early.

“The system creates a web of sensors throughout the forest that communicate through a mesh network and connect back through Spark’s IoT infrastructure, providing real-time intelligence from deep within the forest.”

How the System Detects Wildfire in Its Earliest Stages

Northland’s climate has shifted dramatically in the past decade. Longer dry spells, elevated temperatures, and more frequent high-wind events have increased the region’s vulnerability to fast-moving fire. Recent peat and scrub fires have burned for weeks, emitted enormous volumes of carbon, and stretched limited firefighting resources.

Traditional detection methods like satellite imagery, tower spotters, and public smoke reports tend to identify fires only once they are visible above the canopy. By then, a fire may already be burning at a size and intensity that demands a full emergency response.

Dryad’s Silvanet system challenges that paradigm. By focusing on early chemical indicators rather than visual cues, it brings detection forward into the smouldering phase — the critical early minutes when a small crew can extinguish a potential wildfire quickly and safely.

The Silvanet system relies on a combination of sensors, mesh gateways, and cloud analytics working together as a single early-detection network. Inside the forest, Dryad’s mesh gateways form a dense, self-healing digital network that carries data to the edge of the forest. From there, Spark’s IoT network transports the signals in real time to Dryad’s cloud platform for analysis and alerting.

Each Silvanet sensor is a compact, solar-powered “electronic nose” designed to detect the earliest chemical signals of fire. It continuously samples air chemistry, including hydrogen, carbon monoxide and volatile organic compounds, and pairs those readings with data on temperature, humidity and pressure. Using trained machine-learning models, the system identifies the chemical fingerprint of smouldering material before combustion becomes visible.

Rather than looking for visible smoke or flames, the system learns what ‘normal’ looks like within the forest environment and then identifies unusual changes that may indicate a fire is beginning to smoulder.

Carsten Brinkschulte, Co-founder & CEO of Dryad Networks, explains: “Dryad Silvanet uses ultra-sensitive gas sensors with embedded AI. The gas sensor, along with AI, is able to detect smoke signatures such as volatile organic compounds and sulphur compounds at a smouldering phase, much earlier than any other technology available on the market today.

“By combining those signals with temperature, humidity and pressure data, the system can identify potential wildfire events well before traditional detection methods. The technology continuously learns the environmental conditions within a forest, allowing it to distinguish between normal environmental changes and the chemical signatures associated with an emerging fire.”

Because every forest ecosystem is different, Dryad supports localised training to improve detection accuracy. In Waitangi, FNDC will use samples of local vegetation, including radiata pine needles, mānuka, kānuka, forest litter and humus, to train the AI models specifically for this environment. This plant-material training will occur in a second phase now that the deployment is complete, avoiding delays to the main installation timeline.

The result is a system that learns, adapts, and becomes increasingly aligned with the characteristics of the Waitangi forest.

Once deployed, the sensors establish a baseline understanding of normal environmental conditions within the forest over a period of weeks, allowing the AI to distinguish between everyday environmental variation and the chemical signatures associated with a developing fire.

Spark’s role is central: it provides the national connectivity layer the system depends on, with Spark’s IoT subsidiary, Adroit, managing the local installation and field operations. Together, they ensure the system is delivered with both local expertise and nationwide technical support.

With Spark’s network carrying the data, FNDC receives a continuous flow of environmental intelligence from the heart of the forest to the organisations responsible for monitoring and protecting it.

A Forest with Cultural, Economic, and Environmental Significance

The Waitangi Endowment Forest is primarily a working forest owned by the Crown, with half of its wood-processing income supporting the Waitangi National Trust. When the forest burns, that income burns — along with the forest’s ability to store carbon, provide habitat, and maintain the landscape that supports one of the region’s most important recreational assets.

Nicole Wihongi, Head of Operations & Infrastructure, Waitangi National Trust, says early detection plays a vital role in protecting the whenua. “This whenua surrounds one of the most culturally significant landscapes in Aotearoa — the Waitangi Treaty Grounds. Having that early detection in this part of the forest really benefits the whenua and helps protect it for future generations.”

The adjoining Waitangi Mountain Bike Park (operated by the charity Focus Paihia) also stands to benefit from the new system. With thousands of riders entering the park each year, early detection enables faster evacuation and strengthens the park’s health-and-safety systems.

“The Waitangi Mountain Bike Park is a community-built and community-funded space,” says Tiffany Holland, Park Manager, Waitangi Mountain Bike Park. “It’s taken more than a decade and over $2.5 million in donations and volunteer effort to develop. The fire detection system helps safeguard something our whole region has invested in, both emotionally and financially.”

But above all, the focus is the forest itself. It is the trees that store carbon, the trees that release emissions in a fire, and the trees that take decades to regenerate. The system is first and foremost about safeguarding this landscape and protecting the cultural and environmental significance of Waitangi.

At the same time, the installation creates a valuable opportunity for the wider sector. With the network now in place, Waitangi becomes a live demonstration site where other landowners, foresters, regional authorities, and fire-protection agencies can see the technology operating in real conditions. FNDC plans to host field-day events to support that knowledge-sharing, helping other forests benefit from the lessons learned here.

A Collaboration Grounded in Conservation Values

Installing sensors on conservation land requires careful planning and a deep respect for the environment. FNDC worked closely with the Department of Conservation to secure a 15-year authority permitting the installation and operation of the system within the Waitangi Endowment Forest.

While DOC routinely manages concessions for activity on public land, this approval marks the first time a wildfire detection network of this kind has been authorised for use on a conservation estate in New Zealand.

The approval process involved DOC’s regional offices, the Waitangi National Trust, local hapū, PF Olsen (forest managers), FENZ, and the FNDC Te Hono Māori engagement team. All parties played a role in shaping how the system would be implemented, how kauri-protection protocols would be followed, and how biosecurity concerns (including Argentine ant and kauri dieback) would be managed.

Adroit’s installation approach respects those environmental considerations. Gateways are installed using reversible ground screws to avoid the need for concrete pads. Sensors are mounted using shallow wooden pegs that do not damage timber or interfere with future forest operations. All equipment is solar-powered with supercapacitors rather than batteries, reducing both waste and long-term maintenance needs.

The installation phase involved specialist silviculture contractors led by Craig Zielinski, Adroit engineers installing and aligning gateway infrastructure, Spark technicians verifying IoT coverage and data flow, and Dryad overseeing mesh network commissioning.

A Forest That Can Sense Risk — and a Platform for the Future

With the system now active, Waitangi becomes one of the most advanced wildfire-ready forests in Australasia. The network gives FNDC, Spark, and Dryad real-time situational awareness across hundreds of hectares of forest interior — areas traditionally unseen, unmonitored, and difficult to reach.

A smouldering ember, a lightning strike, or a spark carried on the wind no longer goes unnoticed. The forest is no longer silent until smoke is visible. It is alive with continuous sensing, ready to alert before a fire has a chance to grow.

Because the network is built on Spark’s IoT infrastructure, the data is secure, immediate, and transported reliably across a national carrier network designed for mission-critical applications. This continuity is crucial in remote or difficult terrain, where even small delays can change the trajectory of a developing fire.

The system also creates a pathway for additional IoT capabilities. Soil-moisture sensors, biodiversity monitors, rainfall gauges, illegal-access detection, carbon-flux monitoring, and visitor-safety devices could all integrate into the same network over time.

With mesh gateways and Spark connectivity now extending across the forest, the digital backbone is in place to support conservation, land management, and climate adaptation well into the future.

As Steve Cooney, Technology Specialist – Research and Innovation, Far North District Council, notes, the goal is not simply to install a new technology — it is to create a model others can learn from. Field-day events will give foresters and regional authorities the opportunity to see the system operating in real conditions and consider how similar approaches might benefit their own forests.

“Early detection fundamentally changes response. This technology enables early detection of wildfires, which allows an early response, protecting lives and the livelihoods tied to the tourism economy the forest supports,” Cooney says.

For FNDC, Spark, Adroit, and Dryad, this project demonstrates what a connected forest can be — awake, listening, and ready to alert, protecting itself, its visitors, and the wider community that depends on it.