How scientists are using Artificial Intelligence to predict wildfires

The LightningCast AI model has been a valuable tool for scientists in predicting wildland fire incidents since 2021. Forecasting fire weather is a complex task, as forecasters encounter various challenges even before a wildfire starts and while it is ongoing. Wildfires can erupt at any moment and in any location, making it crucial to consider numerous factors to estimate their speed of spread and the extent of their impact on our lives and communities.

Last week, Heath Hockenberry, the National Fire Weather Program Manager at the National Oceanic and Atmospheric Association's National Weather Service, highlighted the major obstacles faced in the field of forecasting, which include dealing with complex terrain, conducting a 'fuels assessment,' and effectively communicating the risk involved. Hockenberry explained that complex terrain encompasses various geographical factors such as steep slopes and intersecting valleys, which lead to localized weather fluctuations. To accurately predict weather patterns, meteorologists have to continuously interpret and enhance intricate weather models to account for the influence of mountains and valleys on wind patterns and precipitation.

Weather plays a crucial role in the development and lifespan of wildfires, particularly about lightning. The numerous lightning strikes that occur across the United States each year not only result in hundreds of injuries and nearly two dozen fatalities, but they can also instantly trigger wildfires. This dangerous lightning becomes particularly threatening when there is no accompanying rain, allowing a single strike to ignite a fully-fledged fire. Under favorable conditions, these fires can rapidly spread, posing a sudden and severe threat to communities. In some cases, there may be little to no warning before the fire's onset.

To aid forecasters in predicting these disasters, artificial intelligence has been integrated into the process, as human-induced climate change continues to exacerbate the situation.

Hockenberry mentioned that one of the most beneficial advancements that AI can offer is aiding humans in differentiating between the ordinary and the extraordinary. He pointed out that annually, the occurrence of wildfires surpasses the number of tornadoes or hurricanes. With the assistance of AI and other machine learning technologies, it is highly probable that the vast number of fires can be narrowed down to identify the ones that pose the greatest danger to our country.

AI has become increasingly prevalent within the organization, with its utilization in various areas. In the past, it has proven useful in predicting severe weather and monitoring hurricane development, as well as identifying volcanic eruptions and aiding the aviation industry in monitoring cloud conditions. A notable example of this is the adoption of ProbSevere, a mechanism employed by forecasters at the National Weather Service (NWS). This mechanism allows for greater lead time in anticipation of severe weather events by closely monitoring storm development and issuing both severe thunderstorm and tornado warnings. Building upon the success of such applications, the LightningCast AI model, developed by John Cintineo at the University of Wisconsin/Cooperative Institute for Meteorological Satellite Studies (CIMSS), was tested in 2021. This model aims to enhance fire weather forecasts by providing easily understandable and consistently accurate information.

According to Mike Pavolonis, a physical scientist with the NOAA/NESDIS Center for Satellite Applications and Research, AI is the use of technology to automate intellectual tasks that are typically carried out by humans. While humans are skilled at extracting information from satellite imagery, they can only analyze a small portion of the vast amount of environmental data available. Therefore, the automation of these tasks is crucial in order to fully utilize environmental data sources like satellites. One example of this is LightningCast, which has evolved and is now regularly used by NWS forecasters for decision support, aviation forecasting, and more recently, predicting thunderstorms at wildland fire incidents. Thunderstorms pose a significant hazard for fire crews, and predicting such storms can be challenging.

The AI model works by leveraging data from NOAA's GOES-R satellites. Every day, it processes data from over 6,600 images captured by the Geostationary Lightning Mapper and the Advanced Baseline Imager instruments. Through its trained algorithm, the AI model is able to identify complex patterns and predict the areas with the highest probability of lightning strikes within the next hour. It achieves this by rapidly generating maps in just a matter of seconds.

Researchers and forecasters are excited about the future integration of AI in research and forecasting. With support from the Bipartisan Infrastructure Law and collaboration with NESDIS Cooperative Institutes, there are ongoing efforts to combine AI with satellite and environmental data sources. The goal is to develop a new algorithm that can detect fires early, predict their behavior and spread. NESDIS is currently testing a new AI algorithm as part of the Next Generation Fire System (NGFS), which can be applied to various satellites and is specifically designed to identify fires more quickly than existing satellite methods. The NGFS also has the capability to automatically track fires, enabling continuous monitoring of their intensity and smoke production. These advancements will be evaluated by operational users during upcoming experiments conducted by the NOAA Fire Weather Testbed.

Follow us for more updates: PRIMAFELICITAS

If you need help with your innovation, contact us:

?? Book a Meeting: https://bit.ly/pf-call

?? Share a project brief: https://bit.ly/pf-web3

?? Website: https://lnkd.in/dXQ4hQFP