Introduction
Tools based on artificial intelligence (AI) are increasingly being used to predict the behavior of forest fires. It allows the fire services to control the situation better and minimize potential damage to the environment and people. However, it is important that such technologies do not replace people but help them, experts insist. Sensors for detecting forest fires have their use limitations, which will be difficult to function without machine learning fully.
Disadvantages of Satellite Sensors
As for the disadvantages of satellite monitoring, experts note a large area of the minimally detectable fire source, ranging from 1 to 50 hectares (Jazebi, De Leon, and Nelson, 2019, p. 434). Additionally, there is a low frequency of receiving data (several times a day) and a strong influence of weather conditions (Pooley, 2022, p. 7). In windy conditions, a delay (4-6 hours) in detecting even a small fire can lead to serious consequences and increase the cost of its elimination.
Machine Learning as a Way to Improve the Operation of Monitoring Sensors
However, despite all the disadvantages, satellite monitoring is necessary in the case of large forest areas, and there is no possibility of monitoring by other means. The very cost of monitoring using sensors is relatively low, and proper machine learning will be effective (Musinsky et al., 2018 p. 92). Experts emphasize that such technologies work best with people who have information about a particular area (Jain et al., 2020, p. 485). Therefore, they recommend that fire services and residents prepare a special map before the start of the fire season and mark on it the places that need to be protected from fires “at any cost.” Algorithms that work based on machine learning analyze data such as the distance from the fire to the road, terrain features, combustible materials on the ground, and characteristics of other fires in the same area (Tabor and Holland, 2021, p. 11). Then they determine the places where firefighting efforts will be effective and where it is not worth spending resources on them. Thanks to this, the sensor system will be able to work efficiently and minimize the consequences of a fire of any force.
References List
Jain, P. et al. (2020). ‘A review of machine learning applications in wildfire science and management.’ Environmental Reviews, 28(4), pp. 478-505.
Jazebi, S., De Leon, F. and Nelson, A. (2019). ‘Review of wildfire management techniques—Part I: Causes, prevention, detection, suppression, and data analytics.’ IEEE Transactions on Power Delivery, 35(1), pp. 430-439.
Musinsky, J. et al. (2018). ‘Conservation impacts of a near real‐time forest monitoring and alert system for the tropics.’ Remote Sensing in Ecology and Conservation, 4(3), pp. 189-196.
Pooley, S. (2022). ‘A historical perspective on fire research in East and Southern African grasslands and savannas.’ African Journal of Range & Forage Science, 39(1), pp. 1-15.
Tabor, K. M., and Holland, M. B. (2021). ‘Opportunities for improving conservation early warning and alert systems.’ Remote Sensing in Ecology and Conservation, 7(1), pp. 7-17.