The controversial conviction of six Italian scientists found guilty of manslaughter because of how they assessed and communicated risk before the L'Aquila earthquake of 6 April 2009 set a worrying precedent for science’s role in risk analysis of crisis situations. But, there are current examples of how cutting-edge science and technology are saving lives both in natural and manmade catastrophes, such as recent flooding in Chad and conflict in Syria. Geographic or geospatial information science is a discipline that applies the use of geographic information systems (GIS) to capture, manage, and analyze all types of geographical data. Human GIS analysts act as emergency managers to rapidly assess the latest satellite data and give accurate visually-corroborated updates of catastrophes to relief efforts.
When a catastrophe hits, a lot of confusion arises, with families displaced and local buildings and infrastructures disrupted or even destroyed. Relief agencies and non-governmental organizations (NGOs) on the ground need fast and reliable data concerning where people are and what state local infrastructure is in.
They need to make a call. This call is typically answered by the Operational Satellite Applications Programme (UNOSAT). Satellite data is crucial to combat natural disasters; take the devastating superstorm Sandy for example, which hit the US east coast last week. Satellite data was crucial for life-saving crisis maps to help the public stay ahead of the storm. Just last week, UNOSAT was activated to provide satellite image analysis of floods caused by Hurricane Sandy to Grand'Anse in Haiti.
The UNOSAT programme, supported by the United Nations Institute for Training and Research, has been physically based at CERN in Geneva, Switzerland, since 2001, in a fairly inconspicuous building. Apart from a faded UN flag outside its front door, there is little to give its presence away. This may seem like an unlikely place for a global disaster rapid-response team, but UNOSAT’s reach is wide.
When floods hit the Republic of Chad in September and October this year, thousands of houses and crop fields were destroyed, and thousands of people affected. GIS data was instrumental in the relief effort. For example, workers needed to assess an area beyond a road which was hit by massive flooding, several meters high. Questions, such as ‘is it safe to cross?’ and ‘are there displaced people on the other side?’, can only be answered in a very short time by GIS analysis of satellite imagery, which gives an accurate bird’s-eye view of the situation.
According to one of the GIS analysts at UNOSAT, Wendi Pedersen, her job requires her to respond very fast, act as a problem solver, and think on her feet.
“When a disaster occurs, UNOSAT may be activated by another UN agency or a humanitarian non-government organization. We request the International Charter on Space and Major Disasters to activate over the event. The charter is a network of agencies who have agreed to collaborate in tasking imagery to aid in disaster response and humanitarian relief,” says Pedersen.
“The Charter provides us with satellite data for free and this data is then analyzed in a rapid-response effort here at CERN by UNOSAT specialists. The result is rapid and valuable information on the disaster situation to support humanitarian relief operations in the field. This was done for the flood in Chad.”
All information produced by UNOSAT’s Humanitarian Rapid Mapping Service is provided at no cost to the humanitarian community because of a group of donor countries. “We work with a fantastic network of resources and partners, which helps provide valuable information on disasters and crises quickly,”says Pedersen.
Once the satellite images are downloaded via FTP by the UNOSAT team, the GIS analysis begins. UNOSAT compares archived post-crisis satellite images of a disaster event with various processing procedures, depending on the type of event and type of imagery. “For each image type received it is best to have an archive image of similar resolution and sensor used to perform an optimal comparison of signatures seen in the imagery,” says Pedersen.
These images are then stored on servers in CERN’s IT department, which provides 24-hour, 7-days-a-week support. “You wouldn’t believe how vital CERN’s IT support is for the work we do; much time is saved by being able to provide everyone with quick and easy access to our internal data,” says Harry Kendall, who works on IT and web support for the team. To date, UNOSAT has a catalogue of more than 10 terabytes of GIS data and images stored at CERN.
Citizen scientists can be GIS analysts too
One of UNOSAT’s latest projects is to provide mapping and technology support to ForestWatchers, which is an initiative aimed to curb the rapid decline of tropical forests globally, by enabling the public to monitor any selected patches of forests, almost in real-time, using a desktop computer, notebook, tablet, or smart phone connected to the web.
Research is being carried out to help volunteers create deforestation maps, effectively enabling members of the public to become GIS analysts. Partners of ForestWatchers include the Brazilian National Institute for Space Research (INPE), the Federal University of São Paulo, Brazil,and the Citizen Cyberscience Center. Daniel Lombraña of the Citizen Cyberscience Center is developing an open-source platform that will facilitate image classification, called PyBossa, as well as web applications which will allow volunteers to review deforestation directly from their web browsers.
On 21 November 2012, a presentation of results will be held at the United Nations Office, Palais des Nations, in Geneva, Switzerland, on work done in Brazil with the INPE and the Permanent Mission of Brazil, hopefully encouraging expansion of the initiative to other forests and users around the world.
“GIS analysis doesn’t just capture the impact of a disaster,but it enables us to tell a story of what’s happening,” says Pedersen. “We can confirm reports from the field, such as the location of flooded areas, landslides, infrastructure affected, and to what degree. In other cases, like in Libya and Somalia, we can confirm the location and development of IDP (internally displaced people) camps. We can say to our colleagues on the ground where people are moving to, for example.”
To get to this level of clarity, UNOSAT analysts like Pedersen use GIS tools such as ESRI’s ArcMap to do various types of analysis of the satellite imagery. Classification analysis identifies different shades of color in optical images to differentiate trees, floods, and people. Also, analysis of radar imagery, which can penetrate cloud coverage, may show areas where light is reflected back, which could signal flooding or standing bodies of water.
Pedersen says:“We extract the information we need from the satellite image, filter it, and clean it up as much as possible. We apply raster processing [to turn vector-based images into pixel-based ones] for color analysis. Images can be difficult to analyze; for example, if there’re lots of tress flood extent can be hidden.”
The final process is to turn the raster – or bitmap – image back into a vector-based image, with added annotations, and the ability to zoom-in or scale-out with no loss of quality. This is done by uploading and processing this file in a virtual machine. Virtual machines enable many different images to be processed simultaneously, and are cheaper and more scalable than having physical servers.
“The virtual machines also give us cross-compatibility with the ArcMap software. Various GIS analysts and I work with different versions of the software. Some people use version 9.3, but certain analysis coding scripts may only run in version 10.1,” says Pedersen.
This is analogous to working with a Microsoft Word 2010 document that is backward compatible in Microsoft Word 2003, without add-ons. The CERN virtual machines enable UNOSAT’s GIS analysts to work in one software environment, saving time,which could result in lives being saved thousands of kilometers away. This entire process, including UNOSAT’s feedback, can all happen in the space of a working day.