As Temperatures Skyrocket, Sensors and AI Help Fight Wildfires
With wildfires causing historic levels of damage, developers are enlisting sensors and artificial intelligence (AI) to help fight, mitigate, and prevent these catastrophes.
Sporadic wildfires are one of the main results of climate change. What was once a 2-3 month battle each year has turned into a year-round issue for firefighters and first responders. In a research study composed by the Congressional Research Service (CRS), approximately 7.1 million acres were burned by wildfires in 2021.
Now several developers are taking action by creating remote sensors and implementing artificial intelligence (AI) in order to improve fire mitigation, prevention, and defensive measures for wildfires.
Improving Current Wildfire Procedures
In this article, we will dive into the latest sensors that can be used to detect the beginning stages of a wildfire under large tree canopies.
Used once each nightfall, drones are the safer tool for firefighters. In dense forests, helicopters and small planes are not the best options. Image used courtesy of ORNL
Electrical power infrastructures such as substations or distribution plants have an extensive array of sensors and relays to determine if an electrical arc or danger is imminent. But outside of these infrastructures, surrounding dense forests can have little to no forms of protection against wildfires.
Recently, the Department of Energy’s Oak Ridge National Laboratory (ORNL) began using sensors, drones, and machine learning to prevent fires and reduce potential damage to electric grids.
Bosch Sensortec’s Fight Against Climate Change
To fight wildfires, what's ideal is a sensor that can be easily integrated and monitored by firefighters. Along those lines, the Bosch Sensortec BME688 is a 4-in-1 environmental sensor that can gather readings of the air quality to determine changes in pressure, temperature, humidity, and gas mixtures.
The BME688 is equipped with AI-based studio software that allows users to customize an AI model with machine learning algorithms to decipher if increases in temperature, decreases in humidity, and detection of carbon monoxide are present in the area.
A network of BME688s can be mounted on trees to detect the early signs of a wildfire while only consuming 0.18 mAh for every 5-scans. Image used courtesy of Bosch Sensortec
The BME688 offers a gas scanner in addition to the 4 sensors. Similar to other AI models, the BME688 can be given several different samples of gas compounds to create a library on the studio software, this data will run as a machine learning (ML) algorithm to compare to the field gas.
With the data from each sensor, the BME688 is well suited for obtaining the early stages of a wildfire to quickly alert first responders. More information can be found in the BME688’s datasheet.
Taoglas’ Antenna Boosts Comms for LADSensors
Aside from gas sensors, there are other devices that gather critical data such as wind speed, humidity, and CO2 levels. A popular sensor platform that can be used for each of those categories is LADSensors, made by a company of the same name.
LADSensors have been used for detecting changes in temperature, CO2 levels, humidity, and wind speed. The issue with LADSensors is that gathering these environmental changes in a timely manner experiences long-range communication failure due to the dense forest.
The LADSensor platform is improving its long-range comms by integrating LoRa technology and an RF antenna from Taoglas called Barracuda. The Barracuda can establish a 360-degree horizontal grid of the surrounding area and receive data from multiple sensors simultaneously, says Taoglas.
Taoglas’s Barracuda is a collinear dipole design that provides high directional beams with strong reliability to ensure long-distance communication. Image used courtesy of Taoglas
Taoglas has crafted an extensive portfolio of antennas, which are being used to improve the long-distance communication for LADSensors. According to Taoglas, João Ladeira, the founder of LADSensors, has selected a specific Barracuda-brand UV-resistant 6 dBi fiberglass outdoor omnidirectional antenna to be coupled with the LADSensor.
Firefighters can easily layout this LADSensors arrangement by first positioning the sensors 15 km apart which will then connect wirelessly through the LoRa channels to transmit readings to the Barracuda antenna.
Remote Sensing Wildfires with IoT Sensors
Because most fires will begin in difficult-to-reach woodlands, beyond the line of sight, it is important for researchers to find reliable remote sensors that can communicate wirelessly over long distances.
A promising addition to preventative wildfire systems is remote IoT sensors. IoT sensors can detect spikes in CO2 levels that can be transmitted via satellite to cover vast areas. Recently, Milesight created the EM500-CO2, a 4-in-1 CO2 sensor that can be used in harsh environments and easily transmit data through a LoRaWAN protocol. Milesight IoT is one of the leading developers in 5G, AI, IoT, and LoRaWAN technology.
The EM500-CO2 is a 4-in-1 CO2 sensor designed to be used in harsh environments. Image used courtesy of Milesight
LoRaWAN, long range with low power wide area, describes a communication protocol and system architecture. These IoT sensors can leverage LoRa technology over a wide area of network protocols with low power consumption. LoRaWAN allows for the sensor to monitor CO2 levels every 15 minutes and then easily transmit the data via satellite to a cloud platform.
A unique feature the EM500-CO2 holds is the ability to measure barometric pressure. Barometric pressure, also called atmospheric pressure, is a term used by meteorologists to describe when inclement weather is approaching. Weather is said to worsen when there is a decrease in barometric pressure which could then become fuel to rapidly spread a wildfire.
Another form of remote sensing is being introduced by L3Harris which will integrate deep learning into wildfire prevention technology. L3Harris’ Helios cameras can establish on-the-ground weather analytics to anticipate the impact of inclement weather. Several of these large terrestrial cameras can be placed around any infrastructure to create a network of fire prevention.
Providing early detection systems for wildfires is still a work in progress but developers like Milesight IoT and L3Harris are putting in the work to integrate AI and satellite communication to bring remote sensing technology to the conversation.
Methods to Improve for Predicting Unplanned Fires
In a study from the Center for Research and Technology Hellas, Information Technologies Institute (CERTH) in Greece, researchers discussed the various methods for improving early wildfire detection.
The study shared some weaknesses for three methods that would benefit from further research: Satellite, UAV, and terrestrial imaging systems.
Through satellite systems, the ability to have Earth observation allows for a larger coverage of rural lands but is known for issues with communication and can potentially lose data during transmittal.
Heavy smoke is an added risk for firefighters using helicopters to fly over rural lands when an uncontrollable fire breaks out. Unmanned aerial vehicles (UAVs) (drones) are a strong candidate for seeking out wildfires but they are hindered by restricted flight time in harsh weather.
The last method that CERTH researchers observed is terrestrial imaging. A common practice for firefighters is to have ground-based cameras and an array of sensors. The disadvantage is the struggle to transmit data back to firefighters. Without a satellite or large watchtower, the data can also be lost in the dense forest.
Moving forward, a combination of remote intelligent sensing, UAVs, and satellite-based systems is likely to provide better coverage and reliability for wildfire prediction and detection.