Natural catastrophes: Systematic research into causes

What conditions are most relevant for natural hazards to become catastrophes? Forensic disaster analysis can provide answers to this question. Supplemented by real-time information from social media platforms, this approach is perfectly suited to the rapid creation of reliable loss estimates in a crisis region.


Forensic investigation of disasters is a relatively young research approach, developed by the international research programme, Integrated Research on Disaster Risk (IRDR). The analysis here extends beyond the natural event itself to uncover the root causes of disasters through in-depth investigations.

Near real-time analysis to estimate losses

The Center for Disaster Management and Risk Reduction Technology (CEDIM), an interdisciplinary research institute founded by the German Research Centre for Geosciences and the Karlsruhe Institute of Technology, has adopted this research approach and added to it a component for near real-time analysis. Using modern monitoring and analysis techniques, the scientists involved focus on the diverse interrelations between technology, people and society. They search for the crucial factors that are most relevant for the extent of damage and use these to derive prevention measures. The near real-time component makes it possible to obtain information on the type, extent and consequences of a catastrophe within hours or days, and to track the course of a catastrophe. This component is important because the flow of information and user interest are normally greatest immediately after a catastrophe occurs. As part of its research activities, CEDIM adopts and applies its own models and tools for rapid loss assessment. In addition, it evaluates information available from the internet via social networks such as Facebook, Twitter, YouTube or Flickr. Each active user of these networks is seen as a mobile virtual sensor. “Social sensors” like these offer decisive advantages over technical sensors, which only provide data for individual points and for just a few measurement parameters. They are mobile, collect a variety of information and can disseminate such information through various channels. One disadvantage with the enormous quantity of information is that the data is frequently of a subjective character and of varying quality. To make the data of practical use for disaster management purposes, CEDIM launched the “Crowdsourcing” project. Its Systematic research into causes goal is to find suitable methods and procedures to filter the appropriate data from the mass of information and assign it to a specific event and particular location.

Sandy was the acid test

The CEDIM interdisciplinary forensic team has already analysed a number of disasters. These include the June 2013 floods in Germany, Cyclone Phailin, which swept across India in October 2013, and Super Typhoon Haiyan, which caused widespread devastation in the Philippines when it made landfall with the highest wind speeds ever recorded. In the latter instance, the CEDIM experts managed to make realistic assessments of the loss and the number of victims within just a few hours. CEDIM was also active with Hurricane Sandy, which moved across the Caribbean to North America in October 2012. The forensic task force started its work immediately after the storm reached the US Eas Coast. The experts used the platform Twitter as an almost real-time component, collecting more than 5 million tweets and storing them in a database. Th information was then filtered on the basis of keywords like hurricane, flood, damage, victim or power outage. About 3% of the tweets included geo-coordinates, which were then used for further evaluations. Since floods in urban areas are difficult to monitor, these eyewitness reports proved a valuable source of information. Tweets like “Sandy floods 63rd Street”, “The Conestoga River still has a couple of feet before reaching the bank” and “Some may not have power but we all have phones connected”, helped give an indication of what was happening on the ground.

Big potential for real-time analysis

A good picture of the nature and scope of the damage thus emerged from the spatial and temporal distribution of the short messages. The example of Sandy illustrated the potential for real-time analysis using the internet and social media. In conjunction with historical loss and event databases and suitable analysis tools, it was possible to make reliable statements on the extent of the loss within a short time. On 7 November 2012, roughly one week after the event, CEDIM presented estimates of direct and indirect losses for the states of Pennsylvania, New Jersey and New York that came very close to the actual figures. They were more precise than figures from professional risk modellers, although the latter were admittedly available somewhat earlier. In the conflict of interests between, on the one hand, obtaining loss information as early as possible, and on the other hand, uncertainty, each user has to weigh up which is more important. With forensic disaster analysis, through the evaluation of big data from social media, it is at least possible to put loss estimates on a broader footing shortly after an event has occurred.