In this project, the University of Alaska Fairbanks (UAF) will collaborate with Argonne National Laboratory (ANL) to develop an undergraduate-focused research training program, named "FireAID," to develop technologies to fight Wildland Fire with Artificial Intelligence and Deep-Learning in Alaska. The overarching research objective of this project is to combine remote sensing with microbial omics to unlock a new frontier in wildfire risk assessment utilizing the power of artificial intelligence (AI), big data analysis, and high-performance computing (HPC). Concurrently, the project’s training objective is to equitably cultivate a next-generation workforce capable of advancing these fields and enhancing diversity in science, technology, engineering, and mathematics, thereby enhancing Alaska's fire resilience.

Wildfires profoundly impact Earth's atmosphere, surface, and subsurface. To comprehensively understand these effects, the project integrates satellite imagery and microbial omics. Large-scale satellite data (e.g., Landsat, Sentinel2, and VIIRS) will be analyzed with deep neural networks to identify pre-fire indicators (vegetation stress, fuel loads) and post-fire impacts (burn severity, vegetation recovery). Meanwhile, soil microbial data from burned, re-burned, and unburned locations will undergo big data analytics, taxonomic classification, and time series analysis to reveal changes in soil microbiomes. Finally, by integrating these findings with advanced large language models, trainees will craft effective prompts using chain-of-thought reasoning to generate actionable knowledge for wildfire management.

FireAID: An Undergraduate Research Training Program to Develop Technologies to Fight Wildland Fire with Artificial Intelligence and Deep-Learning in Alaska

Arghya Kusum Das, University of Alaska Fairbanks (Principal Investigator)

Mario Muscarella, University of Alaska Fairbanks (Co-Investigator)

Santosh Panda, University of Alaska Fairbanks (Co-Investigator)

Orion Lawlor, University of Alaska Fairbanks (Co-Investigator)

Liam Forbes, University of Alaska Fairbanks (Co-Investigator)

Murat Keçeli, Argonne National Laboratory (Co-Investigator)

Tanwi Mallick, Argonne National Laboratory (Co-Investigator)

Wildfires are a natural occurrence in Alaska's boreal forest and tundra ecosystems. However, climate change is increasing the risk of larger, more frequent, and severe wildfires, threatening lives, infrastructure, and resources. As climate change continues to impact Alaska, future fire managers must be proactive in adapting and innovating their strategies to enhance forest resilience and mitigate wildfire severity and intensity, as recommended by the Alaska State Committee of Research.

Driven by this pressing demand, scientists at University of Alaska Fairbanks (UAF) are engaging in dynamic collaboration with scientists at Argonne National Laboratory (ANL) to develop FireAID, an undergraduate-focused research training program to develop technologies to fight Wildland Fire with Artificial Intelligence and Deep-Learning in Alaska. The overarching research objective of this project is to combine remote sensing with microbial omics to unlock a new frontier in wildfire risk assessment utilizing the power of artificial intelligence (AI), big data analysis, and high-performance computing (HPC). Concurrently, our training objective is to equitably cultivate a next-generation workforce capable of advancing these fields and enhancing diversity in science, technology, engineering, and mathematics, thereby enhancing Alaska's fire resilience.

Wildfires profoundly impact Earth's atmosphere, surface, and subsurface. To comprehensively understand these effects, our project integrates satellite imagery and microbial omics. We will analyze large-scale satellite data (e.g., Landsat, Sentinel2, and VIIRS) with deep neural networks to identify pre-fire indicators (vegetation stress, fuel loads) and post-fire impacts (burn severity, vegetation recovery). Meanwhile, soil microbial data from burned, re-burned, and unburned locations will undergo big data analytics, taxonomic classification, and time series analysis to reveal changes in soil microbiomes. Finally, by integrating these findings with advanced large language models, trainees will craft effective prompts using chain-of-thought reasoning to generate actionable knowledge for wildfire management.

This innovative program equips future fire managers with a comprehensive understanding of boreal forest ecosystem dynamics and transformations, encompassing vegetation and microbiome levels, in the context of wildfires. Trainees will master advanced technologies for informed decision-making in wildland fire management, gaining foundational insights into forest changes through AI-driven analytics. With this expertise, Alaska's future fire managers will revolutionize wildland fire combat, driving innovation in fire likelihood prediction, early detection, fire spread forecasting, damage severity analysis, and novel mitigation strategies. By integrating technological expertise with ecological knowledge, they will develop effective solutions to combat wildland fires.