Stories Where Big Data Built Life-Saving Disaster Management Systems

7 min readJan 17, 2022
Ebook Download Link in the end of the article

In 2017, when Hurricane Harvey was on a devastative spree in Texas, disaster recovery organizations trusted census data to deploy emergency evacuation and management programs. The census data revealed that 86% of the population living in the zip code 77011 spoke Spanish. This helped the emergency management team deploy Spanish-speaking evacuation staff and Spanish awareness protocols, resulting in fast and effective intervention.

That’s not it. Data comes in various heroic forms to save lives during disaster management and recovery. Google set another do-good example with the ‘person finder’ app after the 2010 Haiti earthquake. With its huge database of people’s information, Google united the loved ones who got separated during the earthquake. Here, teams from Google and normal citizens uploaded the names of missing persons in the database. Later, distressed family members used this database as a search engine to locate their relatives.

Data and analytics are enabling organizations to offer support relief and response initiatives to disaster-stricken communities. Datasets such as satellite images, when integrated with crowd-sourced mapping tools, can predict calamities. Similarly, drone footage and weather data can significantly strengthen early warning systems.

Emergency Management Organizations Need Data Hungry Geniuses

But exploring risk-mitigating insights from big data isn’t a cakewalk. False datasets and comparisons between them can lead to chaotic decision-making during a catastrophe.

“When there are information voids it can lead to chaos in a crisis.”
Carlos Castillo,
Research Professor at Universitat Pompeu Fabra, Barcelona

Disaster rescuers and responders sit on a pot of golden data but lack analytical skills. In fact, a survey carried out by Every Action states that 90% of nonprofits collate data, but about 50% of them are not sure, how data can help them”.

Who can manage and handle the data, if not the disaster responsive teams?

It is undeniable that data science has immense promise for both government and non-government organizations. However, getting access to specialists who know data better than anyone else is a typical issue for these groups. Processes are frequently bureaucratic, and resources (financial grants) may be insufficient to hire a full-fledged staff. And many businesses are oblivious to the value of data science.

That’s where organizations with a combination of staunch analytical and software development skills and a good Samaritan attitude build systems that use data for good. They build systems on top of cutting-edge technology such as AI, ML, and Deep Learning to identify actionable insights from data for disaster planning and management.

SUNNY LIVES — A Story that Validates the Alliance of Disaster Responders and Data Science Experts

SUNNY LIVES is an AI model that utilizes high-resolution satellite imagery data of cyclone-prone areas. Now, let’s find out how and why it was built!

Sunny Lives: An AI-driven system to predict, prepare, and respond for natural disasters


SEEDS is on a mission to save the lives of the most underprivileged communities who comprise the bottom 1% and are disproportionately impacted by the gruesome natural disasters and pandemics.

About the Sunny Lives AI Solution

SEEDS’ years of good work have generated huge datasets on disasters management and responses. SEEDS shook hands with advanced technology, under Microsoft’s AI for Humanitarian Action program, to make sense out of all the existing complex data. It also gave them an opportunity to compare and visualize insights from more datasets such as satellite imagery and drone footage.

With advanced machine learning models applied to aerial data, they were able to understand how the roofing material of a house can act as a proxy for its socio-economic condition. For example, a family living in a bamboo thatched house would have less capacity to recover and respond to a disaster than a family living under a concrete roof. Such life-saving insights helped them assign risk scores to individual houses at a neighborhood level.

Data of over 50,000 houses were manually tagged and classified under 7 categories based on the material used for their construction.

Sunny Lives, an AI model, uses high-resolution satellite pictures of locations likely to be affected by the cyclone. The experts then identify the most susceptible houses, using modern data analytics and machine learning.

This allows SEEDS volunteers to identify buildings that are most vulnerable to the cyclone and focus their outreach efforts on those groups. SEEDS could accelerate their evacuation plans to reduce the impact of any incoming disaster for the at-risk population. The final ML model could identify roofs with up to 90% accuracy.

Find out more about this initiative from SEEDS, in collaboration with Microsoft & Gramener, to become highly disaster responsive with AI.

Download the case study of how Gramener build a robust AI model to help SEEDS execute disaster management programs.

How Data Brings Out Value-Driven Decisions for Disaster Management Teams

The past decade has seen a significant rise in the application of data and analytics for disaster management. Data is enabling governments and non-government organizations to make decisions to minimize or mitigate the chaotic effects of disasters.

Climate Data for Insightful City Resilience

Weather data is helping governing bodies in climate resilience planning. In fact, the Canadian government is leveraging AI technology to address a host of climate-related hazards, from flooding to earthquakes. They are also examining heatwaves patterns in the city of Calgary to identify the areas most impacted by the Urban Heat Island Effect (UHI). This is enabling game-changing decisions among municipalities across Canada to plan for and mitigate the impact of climate change in their communities.

How climate data can help explore city resilience opportunities

Satellite Mapping in Disaster Response

Drone footage and satellite mapping data are helping disaster response teams make decisions to rescue communities in crisis. Also, with hyper-local data analytics techniques they can locate hotspots that need immediate action. For example, locating the elderly community or identifying hospitals and schools can assist in the development of alarm and response measures, as well as the movement of at-risk populations to safety.

One heartwarming story of mapping technology saving a life came out during Hurricane Harvey. A truck driver was stuck inside his vehicles in about 10 feet of rising waters and had to be rescued by boat. For the responders, it was impossible to know where the ground lay under the flooded surroundings. With mapping technology, the disaster responders could identify the depth of the surrounding waters.

Acting Rapidly When Earthquake Strikes

Seismic sensors are instruments that detect ground vibrations such as those generated by earthquakes, volcanic eruptions, and explosions. The sensors detect primary waves, the earliest seismic sound waves produced by an earthquake. In dealing with the incident, a graphical image created by interpreting the data provided by these sensors is important. In such circumstances, the public can get a proper warning, which can assist in saving people’s lives.

Social Media Data for Disaster Rescue

Social media is one of the most potent data sets for disaster recovery and relief. During natural catastrophes, social media is the first and only data source available. On-the-ground witnesses update it as the event progresses. After the crisis is over, social media posts are analyzed and compared with other data to track response efforts and better prepare for upcoming events.

One of the examples is The Red Cross society helping local agencies track where the citizens took shelter during storms and understand how much time it took them to return to their homes. This empowers the decisions of local agencies to acquire more volunteers, set up shelters in defined vicinities, and improve on their future disaster response and management efforts.

Sorting relevant pieces of information from thousands of tweets is a challenging task. However, automated tools powered by Natural Language Processing (NLP) models can review and categorize the massive volume of tweets.

Survivors also use social media to generate a real-time picture of what is happening by geotagging places or posting time stamps. Social media may warn officials of affected regions, road closures, power outages, and more by providing direct, essential information from the user.

Data Literacy: A Must for Public Organizations to Fight Hazards

Enabling data literacy across every level of public organizations can augment data-driven decision-making. Gartner defines data literacy as,

“The ability to read, write and communicate data in context, including an understanding of data sources and constructs, analytical methods and techniques applied, and the ability to describe the use case, application and resulting value.”

Public agencies and non-profits can build data literacy by learning how to use open data sources to make information easily accessible. Furthermore, they can invest in data and analytics tools that help in disaster preparation and response.

For organizations like SEEDS, which are collecting and revising their own data, the efforts could go to waste if they’re not sharing the information publicly in a consumable form.

The use of data analytics tools for disaster management has risen dramatically, particularly in underprivileged populations that are vulnerable to natural hazards. These groups must now work together, develop better tools, and share data as it is collected.

Download ebook on how organizations from ESG industry can leverage data and build AI solutions to help them fight problems related to disaster management, biodiversity conservation and building city resilience.




Gramener is a design-led data science company that solves complex business problems with compelling data stories using insights and a low-code platform, Gramex.