You are on page 1of 3

Leveraging AI and Big Data to Better Respond to Crises

A partnership between UNICEF and Data for Good at Meta

Background
UNICEF, the United Nations Children's Fund, is an agency of the United Nations responsible for
providing humanitarian and life-saving support and development assistance to children
worldwide. With a mission to ensure that every child has the opportunity to survive, thrive, and
reach their full potential, UNICEF works to provide health care, education, protection, and
support to children in need worldwide.

In recent years, UNICEF has harnessed the power of data to enhance its response to disasters and
crises globally. Leveraging maps from Meta’s Data for Good program, UNICEF has built a series
of tools designed to fill critical data gaps in the initial 72 hours after natural disasters.

The Impact of Data


UNICEF's Frontier Data Technology team in East Asia and Pacific Regional Office has
leveraged the tools provided by Data for Good at Meta to respond to disasters such as the
Tropical Cyclone LOLA in the Pacific and the Mayon volcano eruption in the Philippines. These
datasets have helped establish baselines for populations affected by natural disasters, areas
experiencing high levels of evacuation or population outflows, as well as the network
connectivity of affected areas.

High Resolution Population Density Maps


To establish a baseline for population density in affected regions, UNICEF has created
visualizations using Meta’s AI-powered High Resolution Population Density Maps, which are
built using a combination of artificial intelligence, satellite imagery and census data to provide a
detailed view of where people live in disaster areas. These maps provide crucial demographic
insights, such as the total population in the affected areas of a disaster, as well as sub-populations
and vulnerable groups such as children under five and the elderly. For example, in the immediate
aftermath of the eruption of the Mayon Volcano in the Philippines, UNICEF teams were able to
quickly determine that approximately 1.3 million people were in the affected areas, including an
estimated 145K children under 5, 633K women, and 103k elderly people. This allows teams on
the ground to tailor their response to the needs of the population at hand.
Facebook Population During Crisis
UNICEF has also leveraged the Facebook Population During Crisis dataset, which compares a
subset of the Facebook population in a particular area following a crisis compared to a pre-crisis
baseline period. UNICEF uses these maps to uncover daily population variations and to monitor
the evolution of areas that have seen the largest drops and increases in population in crisis
affected zones.

Facebook Movement During Crisis


Facebook movement datasets have also served as part of UNICEF’s toolkit. These maps show
many Facebook users who have enabled Location Services have moved from one area to another
in the aftermath of a crisis and have enabled UNICEF to know where people are leaving and
where they are headed.

Network Coverage Maps


Additionally, UNICEF has leveraged Data for Good’s Network Coverage Maps that show where
people on Facebook have cellular connectivity to identify areas that have lost access to cellular
networks. For example, during the Mayon Volcanic eruption in the Philippines, UNICEF teams
could pinpoint locations near Cabasan, Misbis and Jovellar that experienced outages in the days
following the crisis. Connectivity dynamics are also monitored within the tool thanks to dataset
daily refresh.
Combining HRSL population, network coverage, population during crisis and movement maps
will help UNICEF and its partners quickly assess the natural disaster’s impact. They will be able
to respond faster with the right services in the right areas - where internally displaced people are.

Conclusion and Next Steps


UNICEF's data-driven approach, as exemplified by the strategic use of Data for Good datasets,
demonstrates a commitment to innovation in humanitarian response. Over the coming months,
the Frontier Data Technology team hopes to improve its tooling, formally incorporate it as a
parallel method to the usual emergency process, and make it widely available. The hope is that
over time, these dynamic datasets will help better tailor the response to nearly all major crises
that UNICEF teams respond to, saving time, money and lives in the process.

UNICEF does not endorse any brand, company, product or service.

You might also like