Sep. 16, 2008 When disaster threatens, the first hours are crucial. European researchers have developed an automated system to provide early detection, forecasting, and warning of natural disasters such as floods and wildfires.
Floods, forest fires, and other natural disasters take lives, harm the environment, and cause billions of euros of damage every year – €50 billion worldwide in 2007 alone.
The EU-funded programme SCIER (Sensor and Computing Infrastructure for Environmental Risks) took on the challenge of developing a state-of-the-art automated system to detect disasters in the making, forecast how an emergency is likely to unfold, alert authorities, and get them the information they need to respond effectively.
“We can provide public authorities with real data and predictions in real time,” says Sotiris Kanellopoulos, the project’s technical coordinator. “So the public services can coordinate their forces and manage the emergency in an efficient way, and people who live close to forests or rivers can protect themselves.”
The first level of the group’s solution is to deploy networks of ground-based sensors such as video cameras, meteorological instruments, and river-level gauges in high-risk areas, especially the “urban-rural interface”, where homes and businesses lie close to undeveloped terrain.
From raw data to realistic forecasts
The ground-based sensors are linked wirelessly into what the researchers call a local area control unit. This level of the system structures and compares the raw data, for example checking to see if a temperature spike at one sensor is matched by similar changes at nearby sensors.
“The system should be able to understand when there is a false measurement,” says Kanellopoulos, “so it can filter out what is unrealistic and not trigger a false alarm.”
When the local area control unit decides a threat is real, it activates the next level of SCIER’s computational armamentarium to forecast how the emergency is likely to develop during the crucial first hours.
“We don’t claim that we can simulate a fire disaster for days,” says Kanellopoulos. “But we can simulate it for the next few hours.”
The researchers have implemented sophisticated mathematical models of how natural disasters unfold. Those models include detailed information about the local geography, plus real-time sensor data concerning wind, rainfall, temperature, and other variables.
They found that, in order to produce meaningful forecasts, they need to generate multiple simulations of a disaster. Only then can their models provide authorities with accurate and useful information, such as where a wildfire is most likely to threaten homes.
“We generate different scenarios using different wind speeds, directions, and other relevant parameters,” says Kanellopoulos. “Then we score each scenario and try to filter out scenarios that are unrealistic.”
The system uses the most likely simulations to generate detailed maps that authorities can use to manage the emergency.
“The simulations are visualised on a reference map, so the public authorities can see in a very direct way what is going to happen in the area for the next two or three hours,” says Kanellopoulos.
Generating these complex simulations in real time demands enormous amounts of computing power. SCIER relies on the GRID to provide that computational clout.
The GRID, sometimes known as the next-generation internet, is a dedicated network that links thousands of computers via a fibre-optic network that is up to 10,000 times faster than the internet. It allows researchers to perform calculations that could not be done otherwise.
“Because we need to run a vast amount of calculations in real time, I don’t believe that a single core computer could compete with the GRID,” says Kanellopoulos.
First field trials
SCIER, funded by the EU’s Sixth Framework Programme for research, already has a functioning trial network in the Czech Republic, aimed at managing floods.
The next trial is taking place this summer near Athens, Greece. It will test a sensor network, local area control unit, and higher-level computational resources for detecting and controlling forest fires.
A third trial is scheduled to take place in France.
Kanellopoulos says that the group’s greatest technical challenge was in combining research and technological capabilities from different areas, for example ways of generating and presenting geographic information using the GRID.
However, he adds, SCIER’s greatest achievement will be seeing the system applied “on the ground” to help authorities protect lives and property from natural disasters.
“It’s the result that is important – an overview of the event so public services can coordinate their forces and manage an emergency in a more efficient way.”
Other social bookmarking and sharing tools:
Note: If no author is given, the source is cited instead.