A UAV based system for real-time near-infrared monitoring of small-scale wildfires

Abstract

Wildfires are a global threat that is becoming more severe and widespread due to climate change. These fires not only pose a significant risk to human life, firefighters, and infrastructure, but also endanger forest resources, increase greenhouse gas emissions, and cause huge economic losses. Several researchers have been working to find dedicated solutions for early wildfire detection, tracking, and firefighting assistance. Traditional methods of fire detection have mainly been from fire lookouts in towers, infrared sensors on elevated platforms, surveillance of fires from aircraft, and remote sensing from satellites. Although these techniques have been proven to work in other areas, they are unsuitable or are limited in performance due to various reasons, e.g., human accuracy, sensor field of view limiting coverage to smaller areas, sensor cost-effectiveness, and re-visit time on a satellite. To counteract the problem, a real-time wildfire monitoring system that can detect small-scale wildfire events and that can be used for tactical forest firefighting operations is proposed. The concept takes advantage of vegetation biomass combustion by-products such as the alkali element Potassium (K) that is emitted at the flaming phase of the fire. The technique is specific to the flaming phase of the fire and is not affected by the fire size. It employs two high-resolution, cost-effective complementary metal-oxidesemiconductors (CMOS) with high quantum efficiency within the near-infrared (NIR) spectrum. The sensor uses ultranarrow-band filtering and target-to-background rationing techniques for the detection of vegetation fires. The system is designed to be self-contained, having its supporting power, compact, and lightweight for easy integration on different types and sizes of unmanned aerial vehicles (UAV) to provide realtime detection and support to firefighters while airborne. UAVs can provide a low-cost alternative for the reduction of fire disasters through early detection, reporting, and real-time support for firefighters. This paper presents the experimental results of an NIR optical sensor mounted on a UAV carrier that was used to collect data while flying at low to 200m above ground at the Centurion Grassland Flying Club. The results provide evidence of the presence of K in small-scale actively burning vegetation fires observed at different angles and detectable from a UAV. The results support the use of NIR sensor payload for the detection of small-scale fires from a UAV platform.