FUEGO ground is camera agnostic!
Fireball’s AI, FUEGO-Intelligence, is camera agnostic and works with nearly any camera system. In California, we are part of the AlertWildfire.org consortium led by Prof Neal Driscoll (Geosciences Research Division at Scripps Institution of Oceanography, University of California, San Diego). Fireball’s role in this collaboration is providing the AI engine that analyses data of the camera/sensor networks HPWREN and ALERTWildfire and fusing this data with fire-related information flowing from satellite and other overhead networks. This is the foundation of a robust automated early wildfire detection and assessment system.
FUEGO-Intelligence AI is elastic, automatically scaling to processing any level of demand. Currently, we analyse nearly one billion images per year from ~800 ground-based cameras detecting fires as early as a minute after they start. Fireball also analyses roughly 500,000 satellite images per year (one every 60 Seconds). In the USA, Fireball’s FUEGO system monitors approximately 123 Million acres California, Nevada, Oregon, Idaho and Washington. From Mexico to Canada, Fireball issues alerts shortly after wild fires ignite.
Camera/Sensor Networks in the USA
Both the HPWREN, as well as the ALERTWilfire network in California were conceived as emergency call confirmation tools. Their purpose was to allow Fire Agencies to assess a fire after an emergency call was received.
Fireball is enhancing the capability of both networks by analysing the camera data with FUEGO Intelligence and fusing this data with near-real-time satellite images, reporting the fires automatically even before the first emergency call is received.
The HPWREN camera Network
The High-Performance Wireless Research and Education Network (HPWREN), a University of California San Diego partnership project led by the San Diego Supercomputer Center and the Scripps Institution of Oceanography’s Institute of Geophysics and Planetary Physics, supports Internet-data applications in the research, education, and public safety realms.
HPWREN is an interdisciplinary and multi-institutional UCSD research and education project, created in 2000 with National Science Foundation funding by Principal Investigator Hans-Werner Braun at the San Diego Supercomputer Center and Co-Principal Investigator Frank Vernon, a seismologist with the Scripps Institution of Oceanography’s Institute of Geophysics and Planetary Physics. The startup and subsequent existence of HPWREN were funded by grants from NSF in 2000, 2004 and 2009. Following generous funding over a time period of approx. 10 years by NSF, the HPWREN user community joined together to sustainably fund the HPWREN Backbone network to support their programmatic needs, while smaller NSF fundings continued for a multi-year transition period. With Hans-Werner Braun having retired in 2016, the project leadership has been transitioned to Frank Vernon.
The network supports a wide range of network application requirements, ranging from the high-volume astronomical data generated by the Palomar Observatory to a steady output of continuous, low-volume traffic from many devices such as earthquake and other environment-observing sensors, which deliver real-time data.
HPWREN includes permanent sites as well as those created temporarily and on short notice, such as firefighter Incident Command Posts (ICPs). HPWREN saw use in several of the major wildfires to hit San Diego County across many years.
The network spans from San Clemente Island in the Pacific Ocean, via the southern California coast to the inland valleys, east toward the mountain elevations of almost 9,000 feet, and the remote desert, reaching almost to the Arizona border. The network’s longest link is 72 miles in distance, from the San Diego Supercomputer Center to San Clemente Island.
The HPWREN Backbone itself operates in FCC-licensed radio spectrum while integrating a system of off-the-shelf radio technology. To increase robustness, large portions of the network utilise a redundant topology to create more routes and to increase bandwidth. Access paths to the backbone utilise license-exempt or FCC-licensed radio links. The wireless link capacity ranges from full-duplex, 200+ Mbps FCC-licensed to license-exempt sensor access links with orders-of-magnitude less capacity, as driven by actual needs.
Both the County of San Diego and San Diego Gas & Electric provided further resources to acquire additional cameras and weather stations for environmental observations for the HPWREN sensor network, increasing its utility for public safety uses. The cameras were installed in strategic locations on remote mountain tops, overlooking vast areas of mountainous brush and chaparral, and are often used by firefighters to confirm the location or status of an active wildfire, as well as by news organisations and the general public. The weather stations can provide real-time, up-to-the-second wind data via user interfaces and support monitoring of highly localised weather conditions in fire-prone regions as well as research into the formation and prediction of wildfires.
With funding from San Diego Gas & Electric and fire fighting agencies, a very significant public safety component was added to HPWREN, specifically to connect firefighting related facilities.
ALERTWildfire has two components:
1. ALERTWildfire Systems, Inc., which is the commercial entity managing the network.
ALERTWildfire Systems LLC is led by Prof Graham Kent, PhD (Director of the Nevada Seismological Laboratory).
2. ALERTWildfire.org, which is led by Prof Neal Driscoll
It is a consortium of four universities the University of Nevada, Reno, the University of California San Diego, the University of California, Berkeley, and the University of Oregon as well as several private companies and government agencies (full list of partners). The consortium provides near real-time images streams from state-of-the-art Axis Pan-Tilt-Zoom (PTZ) cameras to help firefighters locate, confirm, and assess fires after emergency calls are received.
ALERTWildfire is an expansion of the original network, ALERT Tahoe, which was deployed cameras and microwave networks in the Lake Tahoe region. This initial project was funded through the Nevada Seismological Laboratory (NSL) at UNR by the Tahoe Prosperity Center, the Eldorado National Forest, and the USFS Lake Tahoe Basin Management Unit. Expanding into many new cameras and microwave locations, the consortium is now contracted by Sonoma Water, Southern California Edison, Pacific Gas and Electric, CALFIRE, NV Energy, and many counties, including Marin, Sonoma, Contra Costa, San Luis Obispo as well as the US Bureau of Land Management & Forest Service in Nevada, Oregon, Washington and Idaho. Approximately 300 new cameras are installed each year and the build-out in the Western US is expected to reach 1500 cameras.
Fireball’s contract with the consortium is to automate the fire detection on the ALERTWildfire and HPWREN networks using AI and related image analysis plus fusing that ground-based data with satellite and other overhead infrared information. The result of that processing is a robust and sophisticated system of systems that can detect fires automatically as early as one minute after ignition and continue to characterize the fire as it grows.