The "Smart" Way to Prevent Wildfires: An IoT-Based Approach for Early Wildfire Detection

From Greece to Canada to Hawaii, unabating wildfires have devasted communities and cost billions of dollars’ worth of damage to local economies just over the last few weeks. The most recent wildfires in Maui, Hawaii have left over 100 people dead and could cost well over US$5 billion. The increased frequency and intensity of wildfires in recent years is partly due to climate change as higher temperatures and prolonged droughts exacerbate vegetation flammability. In turn, the carbon emissions from wildfires also accelerate climate change. In 2021 alone, wildfires produced 1.76 billion tons of Carbon Dioxide (CO2) emissions.

To combat this vicious cycle, technology startups like Germany-based Dryad Networks and California-based Torch Sensors offer hopeful solutions with early wildfire detection technology. By providing real-time data on crucial parameters like temperature, humidity, and air quality in remote areas in the forest, these startups provide solutions to prevent wildfires from spreading by alerting authorities of small fires before they turn into deadly infernos.

Previously, wildfire detection was largely reliant on optical systems using either cameras, satellite imagery, or ground patrols. The primary challenge with these traditional methods is the slow detection speed (compared to the speed at which these fires spread). With the advent of Internet of Things (IoT) devices and Artificial Intelligence (AI), sensors can detect fires still in the smoldering phase (within the first hour) and instantly alert the authorities to the location of the fire.

Dryad’s early wildfire detection system, Silvanet, uses solar-powered sensors that detect carbon monoxide, hydrogen, and other gases emitted in the early stages of a forest fire. Through built-in Machine Learning (ML) capabilities, the sensor can analyze gas patterns to detect a fire. To connect the IoT system, Dryad provides a long-range (LoRa) wireless network that has been extended with a mesh network architecture. The LoRaWAN standard is ideal for remote locations in forests due to its low-power, long-range offering. Priced at US$50 per sensor, Dryad is forecast to manufacture 230,000 units by the end of 2023.

Torch Sensors, founded in 2020, offers a similar complete early-stage fire detection system with solar-powered sensors that can detect fires as early as within the first 10 minutes. Each device measures thermal, visible, and chemical signs of a fire with multiple infrared cameras, visible cameras, and gas sensors. The device measures different gases and with built-in AI capabilities, it analyzes smoke signatures to detect a potential fire. The device can identify small-sized heat sources with the help of infrared cameras. The sensor device can also perform spectral analysis, which differentiates a fire from other forms of light, such as sunlight, thereby reducing the chances of false alarms. The data from the sensors are sent via radio signals, with one Wi-Fi-connected device per hundreds of offline sensors. The devices connect to each other in a mesh network. The startup is taking preorders for its first batch of 100,000 devices priced at US$299 per sensor (providing up to 10 acres of coverage). To decentralize the warning system, the app allows users to opt into sharing information with other sensors nearby.

In terms of the technology, a key challenge is the ultra-low power design of the sensor devices. The sensor device needs to detect the initial signs of a fire accurately and quickly and run AI software—all with little energy input. It’s important for the device to have the ability to analyze some of the sensor data, as there isn’t enough bandwidth to transmit all the data for analytics in the cloud. For the system to work, the devices also need to be cheaper, as potentially hundreds of thousands of sensors will need to be deployed. Reducing power consumption is important, as it will allow for the size of the solar panel and energy storage to be reduced, which are important price-driving factors for the sensor device. Moving forward, chipsets that allow device optimization with even lower power and costs will significantly drive the expansion of this market.

While both Dryad and Torch have seen success in their testing, it’s unclear how practical these sensor solutions are at scale. From a cost and implementation perspective, it will be challenging covering potentially hundreds of thousands of acres in remote locations. Nonetheless, it is evident that traditional methods of detection are also insufficient. While IoT sensors cannot solve the problem entirely, they certainly have tremendous potential to help mitigate risks, particularly if sensors are strategically deployed in areas like hiking trails, parks, and barbecue spots, as 80% of wildfires are caused because of human actions.

In addition to IoT sensors, other technologies, such as aerial surveillance systems and satellite imagery, should also be deployed, as these technologies can cover inaccessible locations where placing sensors might not be possible. This multi-pronged approach increases the reliability of detecting a fire in its early stages and can enable authorities to make informed and quick decisions based on accurate data to save lives, ecosystems, and economies.