From early detection to rapid response, artificial intelligence is revolutionizing the battle against wildfires. In this article, we explore how these innovative tools are reshaping our approach to wildfire prevention.
Image Credit: Gorodenkoff/Shutterstock.com
The Worsening of Wildfires in an Era of Global Change
Rising global atmospheric temperatures are worsening drought conditions, heatwaves, and disrupting weather patterns, all of which provide favorable conditions for wildfires. Moreover, the spread of urbanized areas has made forest and space management has become a central issue when addressing wildfires. Human activities have, therefore, resulted in more frequent and intense wildfires.
The western United States has been a key region exemplifying the worsening of wildfires. In the summer of 2018, California experienced wildfires exceeding historical precedents. According to an estimate in a study by Wang et al. (2021), the consequences of those wildfires cost USD 27.7 billion in capital losses, 32.2 billion in healthcare costs and 88.6 billion in further indirect losses. The fires that year were the cumulative result of decadal decreases in precipitation and other climate-related changes in the moisture in vegetation.
To address the worsening of wildfires around the world, policies have typically focused on prevention and preparation. From controlled burns and early warning systems to hiring more firefighters, similar solutions are now commonly implemented, particularly in areas of high susceptibility to wildfires. Nevertheless, preventative solutions were unable to tackle some of the worst wildfires recorded in recent years, such as in Australia (2019-2020 see Boer et al. 2020), Greece (Summer 2023), or Morocco (August 2022).
The Role of AI-Based Sensors in Wildfire Management
The rise of digital technologies in recent decades has resulted in the development of numerous tools and techniques, including artificial intelligence (AI). Although it is broadly used, AI technologies typically refer to machine learning algorithms that sense and analyze data in different formats. When linked to online platforms, AI technologies can provide powerful tools that can be applied across sectors from agriculture to the military.
When considering the use of AI technologies in wildfire management, AI-based sensors are being deployed to detect fires. This is the case for a German company, Dryad, which is testing solar-powered sensors linked to large-scale IoT sensor networks in the United States and Canada. The sensors developed by Dryad measure gases, temperature, humidity, and air pressure, and come “with built-in artificial intelligence to reliably detect a fire and avoid false positives” according to the company.
Other AI-based sensors, such as Cal Fire, are deployed across the Sonoma- Lake- Napa area, and rely on camera-based information to visually identify new fire starts. As a result, this AI system detects newly ignited fires to alert firefighters, whereas Dryad relies on environmental sensing technologies. Both methods rely on machine learning to detect anomalies in their environment to rapidly inform the authorities to the presence of a fire.
Once trained using training data, reliable AI technologies and devices can be produced inexpensively and deployed over large areas for prolonged periods of time particularly if they are solar-powered. Networks of sensors can be used in areas susceptible to wildfires or those closer to urbanized areas to prevent or limit impacts for people. However, despite promising opportunities, AI-based technologies possess a number of inherent limitations.
Drawbacks and Potential Solutions
AI-based sensors are being developed primarily as early onset detectors of fires. However, these sensors require monitoring, replacement, and can be destroyed in fires, which, over large areas, can rapidly increase operational costs.
Moreover, to effectively combat threatening wildfires, AI sensors represent only the first layer of detection in what needs to be a multi-layer system of prevention and preparation. AI-based sensors are essentially fire alarms that can alert authorities to the start or spread of fires. Therefore, their most effective application is in remote areas that are difficult to monitor by humans, or areas close to urbanized areas to alert to undetected fires.
It is also important to note that certain wildfires are beneficial. Wildfires occurring in old growth forests are necessary occurrences in forest ecosystem, and such factors would need to be integrated into the planning of AI sensors. If AI sensors are not deployed strategically and since they do not distinguish between small or large fires, alerting to less threatening wildfires may detract funds and attention from more dangerous incidents.
Reliable detection of wildfires requires a combination of technologies and techniques. Sensors, drones, satellite monitoring, and fire lookouts, all offer different advantages and can be used together in a layered detection system. As detection measures, the combined use of different methods can identify fires rapidly and over varying spatial scales, which is crucial in broad regions such as western United States or Canada.
The Future of AI-Based Wildfire Detection
The occurrence of wildfires will likely worsen in the coming decades. A study by Buccholz et al. (2022) indicated that beyond just the fires themselves, wildfire pollution could potentially impact the health of millions of people. As a result, effective strategies to detect and combat wildfires are increasingly important to implement, including the use of AI-based sensors.
In a 2021 study, Pausas and Keeley describe the reasons underlying the onset of wildfires. The authors present how “no single factor produces wildfires; they occur when fire thresholds are crossed”. These factors include drought conditions and vegetation humidity, which can be integrated into environmental sensors deployed within forests and by remote sensing satellites. The opportunities for AI to be used for both these strategies demonstrate how broadly the technology can be implemented.
Nevertheless, wildfire prevention programs are already suffering from budget cuts in many countries. A study by Tymstra et al. (2020) found that wildfire management in Canada needs to implement innovative tools, enhance collaboration, and adjust its strategic outlook, yet it lacks critical funds to make any effective changes. AI sensors may be the future of early wildfire detection and have promising results to alert to dangerous fires, but they cannot prevent or protect from wildfires alone.
References and Further Reading
Boer, M. M., De Dios, V. R., & Bradstock, R. A. (2020). Unprecedented burn area of Australian mega forest fires. Nature Climate Change, 10(3), pp. 171–172. doi.org/10.1038/s41558-020-0716-1
Buchholz, R. R., Park, M., Worden, H. M., Tang, W., Edwards, D. P., Gaubert, B., Deeter, M. N., Sullivan, T. P., Ru, M., Chin, M., Levy, R. C., Zheng, B., & Magzamen, S. (2022). New seasonal pattern of pollution emerges from changing North American wildfires. Nature Communications, 13(1). doi.org/10.1038/s41467-022-29623-8
Dryad Networks GmbH. (n.d.). DRYAD Networks [Video]. Dryad. Available at: https://www.dryad.net/
Pausas, J. G., & Keeley, J. E. (2021). Wildfires and global change. Frontiers in Ecology and the Environment, 19(7), pp. 387–395. doi.org/10.1002/fee.2359
Tymstra, C., Stocks, B. J., Cai, X., & Flannigan, M. D. (2020). Wildfire management in Canada: Review, challenges and opportunities. Progress in Disaster Science, 5, p. 100045. doi.org/10.1016/j.pdisas.2019.100045
Wang, D., Guan, D., Zhu, S., Mac Kinnon, M., Geng, G., Zhang, Q., Zheng, H., Lei, T., Shao, S., Gong, P., & Davis, S. J. (2020). Economic footprint of California wildfires in 2018. Nature Sustainability, 4(3), pp. 252–260. doi.org/10.1038/s41893-020-00646-7