Real-Time Wildfire Detection From Space – A Trade-Off Between Sensor Quality, Physical Limitations and Payload Size

Abstract

Wildfires cause large scale devastation to human settlements and forests every year and their frequency and severity is on the rise. A major reason for this devastation is the significant delay in their detection due to their remote locations in forests. To mitigate this, a constellation of nanosatellites in Low Earth Orbit (LEO) equipped with multi-spectral visible to Infrared (IR) cameras is proposed leveraging the modular and affordable architecture of CubeSats. Coupled with the payload design, meticulously planned constellation and a ground support system, all surface points on the planet will be revisited at least once in an hour. Capturing a surface location with a high resolution in Mid Wavelength Infrared (MWIR) and Long Wavelength Infrared (LWIR) allows a precise estimation of thermal output of the surface. Simulations indicate that a fire of about four hundred square meters can be easily detected from this satellite payload. Through onboard data processing, wildfires can be already detected in space, minimizing bandwidth requirements for real-time alerts. This enables an early wildfire warning within 30 min by utilizing existing satellite internet networks. Additionally, compressed raw images will be transmitted on fixed ground station passes to provide a global thermal data updated every 90 min. The near real-time multi-spectral data provides opportunity for several other applications like weather forecasting besides wildfire detection.

A Wildfire Smoke Detection System Using Unmanned Aerial Vehicle Images Based on the Optimized YOLOv5

“…As of January 2020, over 1100 CubeSats had been successfully launched by various academic institutions and commercial enterprises worldwide. Shah et al [ 90 ] introduced a system comprising a constellation of nanosatellites equipped with multispectral visible-to-infrared cameras and a ground station. This would allow all surface points on the planet to be revisited at least once per hour.…”

A Synergistic Approach to Wildfire Prevention and Management using AI, Machine Learning, and 5G Technology in the United States

“…With satellites, it is possible to detect extensive wildfires using thermal infrared and high-temperature-sensitive short-wave infrared channels [25]. Due to their low cost and large-area repetitive coverage, satellite devices are useful in near-real-time fire detection, monitoring, and assessment of blazed areas [49]. However, despite exhibiting a high spatial resolution, sun-synchronous satellites have a low time resolution therefore making it challenging to detect wildfires in real time.…”

Comparison and Analysis of Three MobileNet-Based Models for Wildfire Detection

“…This method achieves real-time detection of FMC through satellite remote sensing, providing an effective means for early wildfire warning. Shah et al [3] used CubeSats satellites with a modular and cost-effective architecture in Low Earth Orbit (LEO), achieved early detection of wildfires within 30 min using multispectral visible to infrared cameras. Shaik et al [4] classified fuel types using high-resolution images from PRecursore IperSpettrale della Missione Applicativa (PRISMA), demonstrating the effectiveness of the Scott/Burgan fuel model system through ground assessments, satellite measurements, and simulated concentration displays of summer data.…”