Guest Editorial by Sundeep Reddy Mallu, Head of ESG and Analytics at Gramener

The challenges that governments face after a natural disaster are often not only the incident itself but the vulnerability of at-risk populations. Stopping natural disasters may not be in authorities’ control entirely, but minimizing post-disaster recovery costs is an achievable goal.

In the U.S. alone, in 2021, there were 20 natural disaster events with losses exceeding $1 billion each. So, scientists, UN advisors, practitioners, and policymakers worldwide are showing more and more interest in advanced technologies supporting disaster prediction and early warning.

The use of predictive models for weather forecasts is a familiar concept, and there are recent breakthroughs that prove how AI is fast-tracking global weather predictions.

However, the Internet of Things (IoT), artificial intelligence (AI), and machine learning can be an ally in disaster risk analysis and natural hazards prediction, although not fully embraced yet.

AI-based solutions enable governments to accelerate the execution of planned projects and reduce the potential recovery cost from post-natural hazards.

But expect that AI solutions will have an adoption and learning curve for government departments as the outputs from these solutions will have to go through compliance and regulatory approvals.

(Learn More, courtesy of What’s AI, Louis Bouchard and YouTube.) 

Let’s dive in.

Combining Social Media and AI for Faster Disaster Response

Nowadays, social media platforms are a key source of news. Social media users often provide first-hand actionable intelligence during a tragedy, protest, or election.

For example, early during the coronavirus pandemic, researchers with machine-learning programs analyzed social media, web, and news outlets to monitor the spread of COVID-19.

They searched for posts that mentioned Covid-specific symptoms to see where new hotspots arose and discovered early on that younger people were less affected by the disease.

These posts could eventually assist on-the-ground aid workers and rescue teams arriving at crisis points by helping locate victims and direct efforts to the most vulnerable.

AI and machine learning already have the power to boost the benefits of these online channels even further.

They can help sieve through 350,000 tweets per minute and authenticate photos and comments across all platforms, including YouTube, to distinguish between true and false information.

Another example is, when a government minister’s handle gets tweeted or tagged, it creates a large volume of information.

Isolating this noise and categorizing the tweets is essential to address the issues or requests and allocate resources where possible.

Combining satellite imagery data, seismometers, and AI and predictive analytics tools to scan location-tagged social media comments could offer ground-level location information and verify incidents in real-time, too.

The Vrije University Institute for Environmental Studies published a report analyzing how Twitter activity can help detect floods.

(See why BYU IT and Cybersecurity assistant professor Amanda Hughes was awarded a competitive $24,000 National Science Foundation (NSF) RAPID grant. The goal of the project is to better understand the process of real-time decision-making by human volunteers while they rapidly convert social media data into structured codes that machines can understand. That data is used to help emergency responders in disaster relief. Courtesy of BYU Engineering and YouTube.)

Geospatial AI for Disaster Management

Natural disasters are known to affect marginalized communities disproportionately.

This is because most risk information is delivered at a macro-level, covering hundreds of square meters, instead of localizing risk information down to the neighborhood scale.

One non-profit, SEEDS India, helps citizens with disaster recovery models and delivers granular risk scores down to house level through geospatial mapping.

SEEDS India, Gramener, and Microsoft partnered to build a machine-learning model to identify the most at-risk populations. Through analyzing high-resolution satellite imagery, 50,000 houses were manually tagged in a particular region to create training data.

(ClimateChange is a reality that is changing the lives of vulnerable communities. See how SeedsIndia is helping save lives using AI built by Gramener with Microsoft’s AI for Humanitarian Action grant. Courtesy of Microsoft India and YouTube.)

With photos, the algorithm could detect the socio-economic condition and vulnerability of the owner, and roofs were classified under seven categories based on the material used for construction.

The solution utilized high-resolution geospatial imagery, open datasets, and geospatial analytics tools to host and represent the results. It was deployed at scale in Puri during Cyclone Yaas in May 2021 and could predict houses most at risk with 92% accuracy.

Deploying similar solutions would mean that governments and authorities could warn high-risk houses earlier and prioritize who to evacuate with urgency. Residents could take preemptive measures such as evacuating to a safe site.

At the same time, aid workers would know where to deliver supplies or direct relief operations to benefit tens of thousands of people.

Such actions would reduce disaster impact and lower the overall time and cost of post-disaster recovery.

Modeling and Forecasting for Societal Resilience

“Artificial Intelligence (AI) could potentially add US$450 to 500 billion to India’s GDP by 2025,” said Senior Director and Head of Insights at NASSCOM, Achyuta Ghosh.

Economic growth in India’s tech industry could also be leveraged to combat the country’s suffering relating to natural disasters, as one of the most catastrophe-prone countries.

Governments know that AI is critical for increasing productivity across many different industries, but it is also essential for building a robust and resilient society.

In post-pandemic recovery, businesses realize the need to focus on preventing future disasters. Governments and humanitarian agencies are constantly asking themselves:

Response teams are looking for specifics, down to names and addresses so that aid can be mobilized effectively.

Machine learning and AI can help achieve accuracy in forecasting and predicting volcanic activity by delivering volcanic-seismic data and accelerating data acquisition and analysis.

Every volcano behaves differently, but machine learning combined with satellite measurements can consider the heterogeneity.

Juliet Biggs, a volcanologist at the University of Bristol, and her team use machine learning to spot ground movements around volcanoes due to magma shifts. By training an algorithm to sort through the data, volcanologists can more accurately monitor and predict volcanic eruptions, leading to communities being less hard hit by these natural disasters and building resilience.

Predictive models are scalable flood and cyclone prediction solutions, too.

During the 2019 Chikuma River flood in Japan, Feyera Hirpa, a Senior Data Scientist, found that the inundation model matched the actual flood, validating AI’s power for flood prediction.

During cyclone Yaas, SEEDS India reached out to around 1100 families living in Penthakata, Puri, and disseminated cyclone advisories, ensuring timely evacuation of people from vulnerable areas to safer places.

The use of AI has shown tremendous promise. However, there are still challenges. Accessibility and transparency need to be enhanced for the constructive use of AI and governing guidelines and regulations must be defined.

This is how governments will be able to make data from AI-based models translatable and usable for local communities on the other side of the digital divide.

About the Author

Sundeep Reddy Mallu is the Head of ESG and Analytics at Gramener, a data science firm, and Microsoft’s Gold ISV partner. He helps transform the organization through an advisory in building teams and adopting a data culture.

As TA head, he drives the hunt for great data science team players across industries and campuses and attracts them to join Team Gramener.

Related Technologies…

OnSolve a Triple ‘ASTORS’ Awards Winner in 2021 ‘ASTORS’ Awards Program

American Security Today’s Annual ‘ASTORS’ Awards is the preeminent U.S. Homeland Security Awards Program, and now in its Seventh Year, continues to recognize industry leaders of Physical and Border Security, Cybersecurity, Emergency Preparedness – Management and Response, Law Enforcement, First Responders, as well as federal, state and municipal government agencies in the acknowledgment of their outstanding efforts to Keep our Nation Secure.

OnSolve (First of Three)

• Best Machine Learning & Artificial Intelligence (Tie**)

• OnSolve Risk Intelligence

• OnSolve offers a suite of AI-powered risk intelligence, critical communications, and incident management capabilities, so organizations can control the entire critical event management process, keep ahead of change, monitor disasters, and be empowered to make quicker, more accurate decisions during the times that matter most.

• Organizations are able to gain risk intelligence information that allows them to prepare in advance for disasters and ensure business resiliency, send out mass alerts to notify the right people at the right time during a crisis, leverage a mobile incident management platform to retain full control during emergencies, and more.

• Before, during, and after a critical event strikes, organizations now have the power of AI to inform timely and accurate situational awareness, the relevance and speed of leading mass notification services to manage critical communications, and the ability to holistically and seamlessly manage critical events through incident management – all from OnSolve.

• OnSolve risk intelligence technology continues to keep people safe and informed, allowing them to better protect themselves, their employees, and communities.

(Learn how AI-powered OnSolve Risk Intelligence delivers intelligence that is truly actionable so you can make accurate and informed decisions. Courtesy of OnSolve and YouTube.)

OnSolve (Second of Three)

• Best Security Orchestration, Automation & Response

• OnSolve Control Center

• Gain 360-control of critical events. With Risk Intelligence, Critical Communications, and Incident Management capabilities unified in a comprehensive interface, you’ll have real-time critical event management.

• Spot emerging threats early. With OnSolve’s AI-powered engine users can filter through all the noise and delivers accurate, actionable intelligence, so they can launch crisis workflows earlier.

• Ensure the right information gets to the right people at the right time. With a common operating picture, clients can assess the impact and activate relevant teams for immediate resolution.

• Third-party integration enables clients to manage risk without disrupting existing operations and infrastructure.

OnSolve (Third of Three)

• Best Threat Intelligence Solution

• OnSolve Risk Intelligence

• Get Real-Time Visibility into Critical Events.

• View actionable information on an intuitive dashboard, and easily visualize and filter critical event intelligence.

• Dynamic maps of facilities, logistics, and personnel movements provide visibility into an event’s impact on your business or agency.

(Less than half of organizations surveyed by Forrester Consulting agree that risks and business disruption can come from anywhere. Learn the signs that you too are underestimating the risks to your organization. Courtesy of OnSolve and YouTube.)

• *OnSolve was also recognized with Multiple Wins in the 2020, 2019 and 2018 ‘ASTORS’ Awards Programs.

The Annual ‘ASTORS’ Awards highlight the most cutting-edge and forward-thinking security solutions coming onto the market today, to ensure our readers have the information they need to stay ahead of the competition and keep our Nation safe – one facility, street, and city at a time.