Preparing for the Worst with the Best of AI and HPC 

Sponsored Content by Dell EMC | Intel

Local governments seeking smarter ways to prepare for and respond to natural disasters are turning to processes driven by data analytics, artificial intelligence and high performance computing systems.

Natural disasters are tops among the big recurring news stories of the current decade.  From devastating wildfires that wipe out entire communities to hurricanes that demolish coastal towns, it seems the next natural disaster is always just around the corner — and threatening to be worse than the last one.

Take wildfires, for example. In 2018, California suffered its most destructive wildfire season on record. The year’s disasters included the record-setting Mendocino Complex Fire, which burned nearly 460,000 acres, and the Camp Fire, which claimed at least 88 lives, 153,000 acres and 18,800 structures.[1] These deadly wildfires came in the wake of a catastrophic year for hurricanes. Chief among those natural disasters was Hurricane Harvey, which struck Houston and points beyond. It claimed at least 68 lives and caused an estimated $18 billion to $20 billion in damages.¹

These sorts of catastrophes leave people wondering whether natural disasters are more frequent and getting worse in their intensity. And the quick answer to that question is: Yes, that’s the way it is. That’s according to a national climate assessment from the U.S. Global Change Research Program. This assessment found that the United States is experiencing “more frequent and intense extreme weather and climate-related events.”[2]

Findings like these increase the urgency for local governments to develop smarter approaches to disaster preparedness and recovery. And for these initiatives, people are increasingly looking to data analytics and artificial intelligence solutions powered by high performance computing systems.

How AI can help

AI-driven applications, including those based on machine learning and deep learning techniques, can enable smarter approaches to disaster preparedness and emergency-response operations. A report from Prowess Consulting, for example, notes that the opportunities for leveraging machine learning in this space span from prediction and planning to on-the-ground responses.

A few examples from the Prowess report:

• Machine learning can be applied to better predict disasters caused by weather, fire, earthquake or disease.
• Advanced threat modeling and simulations can be used to build better emergency response and disaster-relief plans around logistics for evacuations, relief supplies and medical responses.
• Emergency responders can use AI with image and data processing to pinpoint danger zones and more effectively target their responses.[3]

Let’s look at a few specific examples of AI and data analytics applications in disaster preparedness and recovery.

Improving storm forecasts

One of the ways in which AI could help local governments prepare for natural disasters is through better weather modeling and forecasting.  Work in this direction is under way today at the University of Texas, where researchers are using a powerful new Dell EMC supercomputer, named Frontera, to improve hurricane forecasts.

“We run a model that simulates the storm surge hitting the coast,” Clint Dawson, a UT Austin professor, explains in a CBS Austin news story.  Used in emergency management, “They take our model runs and decide if they need to evacuate.”[4]

Professor Dawson says the Frontera system — the world’s fifth fastest supercomputer, based on TOP500 rankings[5] — allows researchers to create more models with more up-to-date information to make more accurate storm predictions.

“The impact of this machine, really, is to save lives,” he says in the CBS Austin story.

Fighting wildfires

Drawing on the computational power of a massive supercomputer built by Dell EMC, researchers at the San Diego Supercomputer Center (SDSC) have developed WIFIRE, a new AI-driven weapon in the fight against wildfires.

When a fire breaks out, WIFIRE can run many simultaneous simulations on the supercomputer, named Comet, to help emergency personnel understand where the fire is, how fast it is moving and what direction it is heading. In making its predictions, WIFIRE also factors in data from high-resolution satellite imagery and hundreds of remote weather stations. The ultimate goal is to make the direction and speed of a fire known as early as possible to assist in rescue and containment efforts, according to SDSC.[6]

Identifying communities in need

In disaster recovery operations, organizations are increasingly analyzing social media to gain a better picture of what is happening on the ground and where their assistance is most needed. The American Red Cross, for example, is doing this to help response organizations anticipate the needs on the ground during times of disaster.

With funding from Dell Technologies, the Red Cross now operates three Digital Operations Centers where it monitors social media conversations as disasters strike, using real-time data to identify communities in need. When a disaster strikes, the Digital Operations Centers mine data from social media sites to find indicators of the location and extent of damages. The centers can then feed critical details about damaged areas — including addresses and photographs — to disaster response teams. [7]

In another social media initiative, emergency-response organizations interested in gaining real-time insights from the social sphere can now leverage a platform called AIDR, which stands for Artificial Intelligence for Disaster Response. This free, open source software enables organizations to collect and classify social media messages, including tweets and Facebook posts, related to emergencies, disasters and humanitarian crises. AIDR uses human and machine intelligence to automatically tag up to thousands of messages per minute.[8]

Key takeaways

Faced with a growing threat of catastrophic natural disasters, local governments need to develop smarter approaches to disaster preparedness and recovery. Increasingly, these efforts will rely on data analytics and artificial intelligence solutions powered by HPC systems.

The good news: The key technologies are now in place for the development and deployment of these life-saving solutions. Now the challenge is to put these solutions in place today — before the next natural disaster strikes.

To learn more

• Read the report from Prowess, “The Journey to AI in the Public Sector: Dell EMC solutions and Intel technologies accelerate the evolution of analytics and AI for public safety and security.”
• Be part of the Intel® AI Builders ecosystem of industry-leading independent software vendors, system integrators, original equipment manufacturers, and enterprise end users who have a shared mission to accelerate the adoption of artificial intelligence across Intel platforms.
• Visit com/hpc and join the conversation @dellemcservers

[1] Insurance Information Institute, “Facts + Statistics: Wildfires,” accessed August 30, 2019.

[2] U.S. Global Change Research Program, “Fourth National Climate Assessment. Volume II: Impacts, Risks, and Adaptation in the United States,” 2018.

[3] Prowess Consulting, “The Journey to Ai in the Public Sector,” 2019.

[4] CBS Austin, “UT getting new supercomputer, will help better forecast hurricanes,” August 29, 2018.

[5] TOP500.org, TOP500 List – June 2019.

[6] San Diego Supercomputer Center, “Northern CA Wildfires Generate 1.5 Million Views of UC San Diego’s ‘Firemap’ Resource,” October 13, 2017, and NPR, “Supercomputers Assist Firefighters As Wildfires Spread in California,” November 18, 2018..

[7] American Red Cross, via HuffPost, “How 10 Years Have Changed Disaster Preparedness and Response,” August 17, 2016.

[8] AIDR project, “AIDR - Artificial Intelligence for Digital Response,” accessed August 2, 2019.