Digital Disaster Response

How digital technology can help in emergency and catastrophe situations.

Drones deliver data vital to survival following floods, crisis maps help disaster relief workers to communicate, and apps tell people where to find shelter when a hurricane strikes. In the past decade, disaster response has gone digital. But that’s just the beginning: algorithms and software are playing an increasingly important role in recognising and dealing with crises.

In 2007 digital data was used systematically for the first time to document a crisis. Following the elections in Kenya that year, there were violent clashes on the street. To get an overview of the situation, the Kenyan NGO Ushahidi developed an SMS-based service. All Kenyans were asked to report incidences of violence by text message to a central number. These reports were plotted on a digitised map, which illustrated the extent of the violence. Since then the software has been continually developed and made available as open source software. It has been used repeatedly to document other crises and catastrophes.

For instance, in 2010 after the earthquake in Haiti, Ushahidi played an important role in keeping track of developments and analysing the situation in real time. A group of engaged individuals scattered all over the world – networked through organisations such as CrisisCommons – succeeded in combining text messages from those caught up in the disaster with social media updates to create a detailed map of the situation. That meant the relief workers on the ground knew where people were trapped in rubble, where emergency supplies were being distributed, and which streets were still accessible.

Big Crisis Data: Better oversight and analysis of the crisis situation

Digital disaster relief is evolving. When Hurricane Sandy ravaged large areas of the USA in 2012, relief workers were able to make use of apps and crisis maps. And a growing number of apps allow people to send out appeals for help directly. For example, following the Lushan earthquake in 2013, Chinese agency Iyiyun not only developed interactive, user-generated maps which could be uploaded, but also the “signal bomb”. When users activate this app following a catastrophe, calls for help are immediately broadcast over various social media networks. Also in 2013, after Typhoon Haiyan struck the Philippines, drones were deployed for the first time in a crisis situation.

During a catastrophe an incredible amount of data is generated. Aggregating this mass into usable information is a central challenge of digital disaster relief. Patrick Meier from CrisisMappers says: “Haiti was our first battle with big data – what I call big crisis data. We had hundreds of volunteers monitoring social media and news online and we were just completely overwhelmed.”

One example from a study conducted by the Pew Institute shows how important quality control is with large and unverified data sets. During Hurricane Sandy half a million photos were published on Instagram and over 20 million tweets were sent out. But only 34 percent of these contained genuinely useful information.

U-turn by NGOs: From mistrust to serious interest

Because software can’t automatically read photos, people still have to manually describe what they show. Easy-to-use tools such as ImageClicker help with this. Volunteers rank on a scale how bad the destruction is, which helps coordinate response and reconstruction efforts. To better estimate the destruction from Hurricane Sandy volunteers from the organisations Palantir and Team Rubicon combed through satellite photos of the affected areas and tagged them according the extent of the destruction.

Many of the digital disaster relief organisations coordinate under the umbrella organisations Digital Humanitarian Network. The DHN also serves as a contact point for established humanitarian organisations – from the UN-OCHA to the American Red Cross – which can make use of digital supporters. After an initial phase of mistrust, the UN agencies and big NGOs were eventually won over by positive experiences with technology-driven groups, and today they often work hand in hand. The American Red Cross for example has partnered with Dell to open a Digital Operations Center, letting them manage social media communication through more effectively in the event of a catastrophe.

If the government isn't up to it, then we’ll do it ourselves

It’s not just disaster relief professionals who profit from digital media. Citizens can also organise themselves more effectively. People affected by catastrophes are in any case the first line of assistance and responsible for around 90 percent of aid actually delivered. For example, in 2013 in Germany when large parts of the country suffered heavy flooding, thousands of members of the public coordinated their help efforts using Facebook and Google Maps. In publicly open groups like “Fluthilfe Dresden” or “Deggendorf räumt auf” they organised the distribution of sandbags and found emergency accommodation for people driven from their homes.

On the Indonesian island of Java a community living near the highly active volcano Mount Merapi has established a crowdsourced early warning system for eruption. Their main tool for this is Twitter, along with local radio – and they do a better job than the government could. With the hashtag #jalinmerapi the self-organising eruption alert network lets people know about evacuation routes, safe accommodation or supplies of food and water.

This kind of initiative means a power shift which not all aid organisations are in favour of, because the more populations can provide themselves with information, money, clothing, building materials or food, the more likely they are, in so doing, to be competing with what these established aid organisations provide, sometimes even making them superfluous.

Keeping an eye on Twitter and Facebook means you’re quicker to the scene

Authorities and aid agencies are also increasingly using social media to communicate more directly with those caught up in natural disasters. The Philippine government followed a carefully designed social media strategy in the days before Typhoon Pablo struck in 2012. Storm warnings were broadcast by radio, TV and the government’s official Twitter account (@govPH). Tweets with the official hashtag #PabloPH were plotted on a map and were used by the UN-OCHA to speedily identify areas of damage. This communications strategy proved to be extremely successful, by making use of the media channels already very popular with the population. 

The US disaster relief authority FEMA also uses social networks, not only to stay informed about emergencies, but also to disseminate information more effectively amongst those affected. During hurricane season, FEMA response teams are on average 12 to 24 hours quicker on the scene than they were pre-internet. And the UN now takes just two days rather than five to produce a first situation report.

The oracle potential of big data to predict crises

Twitter itself has recognised the potential of its network in crisis situations and in 2013 it launched Twitter Alerts. People can use the service to follow the streams of verified accounts, thereby avoiding the risk of slipping into erroneous messages and rumours.

But even before a disaster strikes, digital media can be of great help. Big data can serve as a seismograph for looming crises such as natural catastrophes or epidemics. The UN organisation Global Pulse for example scrapes data from social networks and mobile phone networks. And the World Meteorological Organzation uses a worldwide network of satellites, weather stations and communication centres to calculate meteorological and seismographic risks and give advanced warning to populations in danger.

A mass of data is automatically generated – but checking it manually is hard work

The new generation of catastrophe tools are based on the premise that we require fast, automated systems to make these masses of data usable. Digital tools such as AIDR (Artificial Intelligence for Disaster Response) or Twitris can categorise online news reports using algorithms and analysis of key terms. Crisis maps are created from the results which map the extent of the destruction and the acuteness of local needs in real time.

But the data needs to be checked over by humans, and thousands of digital disaster relief workers saved, verified, geo-referenced, translated and analysed the crisis data. Standby Task Force, for example, is a community of digitally-connected engaged people from 95 countries where 85 different languages are spoken.

Conclusion

Compared with other areas of social engagement, disaster relief has digitised extremely quickly. That’s down on the one hand to the inherent urgency of crisis situations, which make immediate communication vitally important; and on the other hand to the relatively large budgets that governments and aid agencies preside over. After small initiatives such as Ushahidi had created and tested innovations with great success, the big players overcame their scepticism towards the application of digital technology. Business, academia and civil society collaborate in this field amazingly well. The broader population is also increasingly making use of online platforms and social media, emancipating itself from total reliance on government agencies and NGOs. These flexible networks react incredibly well to catastrophes and increase the resilience of the society affected.