Early Wildfire Detection and Assessment

Currently, detecting wildfires relies on the human eye and they are often reported 15 minutes or more after ignition. In contrast, Fireball’s real-time mapping can detect fires within minutes, allowing for rapid, effective fire suppression, and the protection of the environment, settlements, infrastructure, people and animals.

Fireball.International Pty Ltd offers a unique fire detection, mapping and intelligence system. Fireball’s technology reports wildfires as early as three minutes after ignition using powerful deep machine learning algorithms that can detect and evaluate their growth in real-time.

Proven Technology

On the night of October 23, 2019, a satellite used by Fireball detected the Kincade fire in California 66 seconds after it began due to a fallen power line. The Fireball intelligence system confirmed the fire within three minutes using the ground based ALERT Wildfire sensor network even though it was dark and the fire was in a canyon not directly visible to the sensor. The explosive fire growth was so clear from the data that authorities were immediately alerted and a town 10 km away was evacuated 20 minutes after ignition. The fire department had yet to physically assess the blaze.

Kincade fire

Smoke detection

Fireball uses automated machine-learning-based image analysis to offer what is demonstrably the fastest way to detect even small fires before they become major ones. Real-time image analysis of both small and major fires provides real-time on-screen simultaneous fire intelligence for firefighting teams across different locations.

In San Diego County alone, Fireball reported 494 fires in October and November, 2019. The system detected and confirmed an unauthorised debris burn in the remote mountains of San Diego County, six minutes after it was lit.

Australian Rollout

Fireball’s system provides early fire detection and mapping services protecting infrastructure and populations. Key users may include forestry, fire and emergency services, local and state governments, forest plantations, primary producers, mobile phone tower operators, power companies, and grid operators.

The Fireball intelligence system works with a wide range of sensors, and already makes use of existing ground-based cameras, such as those of the ALERT Wildfire and HPWREN networks and others in California. Fireball can also arrange the procurement and maintenance of sensors where no network exists.

The company is offering a full-service data product where the client maintains full control of the data. The client receives notifications and imagery verifying the alert and also intelligence that documents and maps the fire’s growth in real-time. Our offering is a simple and inexpensive subscription model based on the number of sensors feeding data into our system.

After our success in California and the devastating Australian bushfire season, we were motivated towards initiating an Australian roll-out of Fireball’s technology as soon as possible.

Finally, we note that Fireball.International is working in collaboration with UC Berkeley, Livermore Laboratories, USQ, ANU and others on a project called FUEGO: Fire Urgency Estimator in Geostationary Orbit. FUEGO is the world’s first dedicated real-time bushfire detection satellite technology and is thus being designed and developed in collaboration with an Australian company and universities. The payload is planned to be launched by 2024, riding piggy-back on an Australian telecommunications satellite into a geosynchronous orbit from which an entire Earth hemisphere can be simultaneously monitored for wildfires.

The company is based in Queensland, Australia but also operating in the USA and has recently secured a major contract in California. The effectiveness of the technology has been demonstrated in a recent publication in an independent, peer-reviewed international research journal (Govil, K., Welch, M.L.,Ball, J.T., and Pennypacker, C.R., ​Preliminary Results from a Wildfire Detection System Using Deep Learning on Remote Camera Images​. Remote Sensing, 2020, v12, p166.).