In recent years, the frequency, intensity, and cost of natural disasters have increased, creating widespread challenges for federal and state governments and the private sector. One such challenge is ensuring the resilience of the nation's infrastructure, including the energy grid.
This concern has led scientists at the U.S. Department of Energy's Pacific Northwest National Laboratory (PNNL) to begin researching how artificial intelligence (AI) and machine learning (ML) can be used to predict wildfires and mitigate the impact of other potential natural disasters.
PNNL's groundbreaking work, pioneered by experts like Chief Scientist Andre Coleman, has helped modernize disaster and emergency management by using AI and ML to interpret vast amounts of satellite and geospatial data in near real-time, helping public safety officials, utilities, and land use managers better prepare for impending disasters.
Where it all began
Coleman said PNNL began experimenting in 2014 with a vision to use imagery from satellites, aircraft and drones to quickly assess hazards posed by wildfires, floods, hurricanes and other weather-related disasters and their impacts on critical energy infrastructure.
“The idea from the beginning was, 'Can we use these types of imagery resources quickly to assess a hazard? What is the extent of the hazard? Where do we see the damage caused by the hazard?'” Coleman says. “And then how do we link those damages to critical energy infrastructure?”
To do this initial work, PNNL built machine learning models and tools to “take that imagery and automatically evaluate it,” Coleman said. Then, over time, more machine learning algorithms were built into the models.
current situation
PNNL's initial focus on near real-time response has evolved into proactive risk assessments, enabling emergency management and operations professionals to make decisions focused on reducing risk well before a disaster strikes, such as managing vegetation in and around critical infrastructure or replacing wooden utility poles with steel poles that can withstand more extreme events.
This change in focus has been prompted by the increasing severity of wildfires, particularly the historic fires that burned more than 10 million acres in 2017 and 2020, leading to wildfires being recognized as a national emergency.
“I think people are realizing that fires aren't what they used to be,” Coleman said. “They burn more intensely and they spread faster.”
Advances in satellite technology help drive PNNL's current approach, which combines data from a variety of satellite sensors, including passive and active, to derive actionable insights. Passive sensors capture reflected energy, while active sensors penetrate clouds, storms and smoke to enable continuous monitoring in adverse conditions. By leveraging open-access satellite data and commercially collected imagery, and collaborating with organizations such as NASA, the U.S. Geological Survey and the European Space Agency, PNNL is able to leverage a robust data ecosystem for its analysis.
Additionally, PNNL's development of Rapid Analytics for Disaster Response (RADR) is a key milestone. The platform combines multimodal imagery, AI, and scalable cloud computing (through partnerships with major cloud providers) with infrastructure damage assessment tools that help users better understand the current impacts and risks to infrastructure from wildfires, floods, hurricanes, earthquakes, and more. Today, when a provider collects and makes new imagery available (which typically takes about four to six hours), RADR can retrieve it, process it, and deliver it in seven to 10 minutes.
But new satellite communications and increased automation have reduced this delay to less than an hour, Coleman said. What's more, this significant advancement in data collection and the increase in the amount of available data is something the RADR cloud platform has planned for and is well equipped to keep up with. For example, new planned satellite sensors such as FireSat will provide 20-minute satellite revisits, so generating analytics for these data requires a highly scalable system like RADR.
Additionally, PNNL's research has attracted the attention of a wide range of stakeholders, including the Department of Energy's Office of Cybersecurity, Energy Security and Emergency Preparedness and the Joint Artificial Intelligence Center, which have played key roles in supporting PNNL's efforts. Collaborations with public and private utilities further enhance the research environment and ensure alignment with practical applications and industry needs.
Where are we headed?
Wildfire detection and forecasting is complex, requiring continuous refinement of AI and ML algorithms. PNNL's multidisciplinary team of scientists, developers, and cloud architects remains dedicated to enhancing the system's capabilities. With specialized Earth observation satellites such as NASA's SWOT and NISAR soon to be available, PNNL aims to further expand its forecasting capabilities and reach a wider range of users. “The idea is to continue to add and improve our algorithms, incorporate sensors that come online, continue to mature our capabilities, and try to make this available to as many end users as possible,” Coleman says.
PNNL's efforts are expanding beyond the threat of natural disasters to a more holistic approach to protecting critical infrastructure, including the increasing number of physical and cyber attacks on critical infrastructure. By integrating environmental, cyber and physical security considerations, PNNL aims to build a more resilient energy grid that can withstand a range of challenges.
“A big part of our lab's role is to look at all of those aspects and say, 'How do we protect the grid? How do we make it more resilient as our reliance on energy continues to grow?'” Coleman says. “Through the different teams that are working on that work, we have an opportunity to collaborate and think about how we can move forward and build a better, more resilient system.”
PNNL's continued innovation and collaboration promises to result in a more resilient energy infrastructure in the future. By harnessing the power of AI and ML, PNNL is poised to address evolving threats posed by natural disasters and strengthen the resilience of the grid for future generations.
This report was produced by Scoop News Group for FedScoop and is part of a series on innovation in government. Microsoft FederationFor more information about Microsoft AI for government, please visit: Sign up here Receive news and updates on how advanced AI is empowering your organization.
