Disaster prediction - can we anticipate deadly natural events?

As communities across the windswept American mid-west prairies come to terms with the latest spate of deadly twisters, researchers worldwide are again asking the billion dollar question; how can we make Tornado Alley safer?

On 20 May this year, a massive and powerful tornado hit Moore, Oklahoma. With peak winds of 340km/h, the 2km wide tornado tore through the heavily populated town on a 27km long path across the Oklahoma City region, killing 24 people and ripping down 13,000 homes, schools, farms and businesses. Water supplies were cut, more than 61,500 power outages were registered and Oklahoma Department for Insurance estimates claims will top $1bn.

Amid reports that meteorologists calculated that the energy released during this storm exceeded 600 times the power of the Hiroshima bomb, the grim reality is the Moore tornado is only part of the picture. This twister was one of 53 in a large, violent weather system that swept across the midwestern United States and lower Great Plains from 18 to 21 May, killing two more people. What's more, this spate of tornadoes came only a day after a 23-strong outbreak that spanned Texas, Louisiana and Alabama, killing six people.

Set these against a back-drop of hurricanes, earthquakes, floods, and tsunamis that take place worldwide, year in, year out, and one fact becomes very clear; each and every extreme weather event is so very difficult to predict.

What are tornadoes...?

Tornadoes are violently rotating columns of air, often developed from a class of thunderstorms known as supercells. During these storms, cold and warm air combine to create updrafts of warm air and down-drafts of cold air.

Tornadoes form as air starts to spin due to winds at different heights blowing at different speeds, creating wind shear. This shear causes the air to spin horizontally, and if this gets caught in the supercell updraft, it speeds up and tilts towards the ground.

As tornadoes develop, funnel-shaped clouds extend from the base; if the cloud then touches the ground, this is what forms the tornado. This process can take place within minutes, which, of course, is the problem.

"We are a long way from having sufficient computer resources to routinely run models complex enough to resolve the small processes within thunderstorms," explains Dr Andrew Barrett, a severe storm researcher and meteorologist from the University of Reading. "To predict the location of tornadoes, we need a computer model complex enough to resolve these processes in the storm. These models exist... but the computational and financial cost of running them means they cannot be run routinely or over a large area."

Arguably, computer resources will eventually meet this need, but issues exist around the weather observations, or storm data, that feed into forecasting models. Ground weather stations, used to measure temperature, humidity and wind speeds, are up to hundreds of kilometres apart, so small scale variations are difficult to detect. What's more, weather satellites and Doppler radar don't pick up crucial features such as pockets of strong wind and dry air that circulate several miles above the ground in thunderstorms and can trigger tornadoes.

"A model needs information on these scales to run and will fill in unknown data as best it can," says Dr Barrett. "Inevitably the model data will then contain errors and because the atmosphere is a chaotic system, these will grow larger, further into a forecast. Any predicted thunderstorm could be in the wrong place and at the wrong time."

Storm data

Thankfully, millions of pounds have already been poured into researching and developing new instrumentation to sample storm data. One key example is a distributed radar system, pioneered by Professor David McLaughlin, director of the University of Massachusetts Engineering Research Centre for Collaborative Adaptive Sensing of the Atmosphere (CASA).

A dense network of, say, 12 small, solar-powered radar are deployed on rooftops and can monitor a London-sized region, detecting the hazardous weather closer to the ground that Doppler radar systems miss.

"If data from a big radar tells us that a thunderstorm has tornadic potential, we put our radar in this region to probe the situation down at lower levels. Once we see rotating air flow, we allocate more resources," explains McLaughlin.

The researchers' system provides minute by minute storm updates - traditional radar typically updates every five minutes – and improves today's average 16-minute warning times by 25 per cent. Prof McLaughlin adds: "Our radar confirms whether or not a tornado is on the ground. We can even follow vortices right down to individual streets."

So why wasn't such as system in place at Moore? Cost can be an issue, but as as Prof McLaughlin highlights, in this instance the system probably wouldn't have helped. Moore is located at the heart of Tornado Alley, with numerous large radar nearby.

"This particular event was viewed by radars with as much resolution as is possible today," he says. "I expect we will be concluding that the tornado was well warned but the sheltering options in that region were inadequate to protect everyone."

Clearly not even multi-million dollar advances in prediction methods can forestall some extreme weather events. As the US tornado season ends, the Atlantic Hurricane season starts with the US National Oceanographic and Atmospheric Administration predicting high activity.

Extreme weather events

Despite Hurricane Katrina and Sandy, the Joplin and Moore Tornadoes, Mumbai floods and more, researchers still have not detected an increase in extreme weather events. But they expect it; and they're putting it down to climate change.

The Intergovernmental Panel of Climate Change (IPCC) recently published 'Managing the risks of extreme events and disasters to advance adaptation to climate change'. An increase in 'heavy precipitation events' and more intense and longer droughts were noted, with a poleward shift in northern and southern extra-tropical storm tracks deemed likely.

More recent forecasts from Dr Reindert Haarsma from the Royal Netherlands Meteorological Institute, support this. In his recent Geophysical Research Letters paper, 'More hurricanes to hit western Europe due to global warming', the meteorologist describes how high-resolution climate models show warming effects will boost hurricane-force storms over Norway, the North Sea, western UK and the Gulf of Biscay from August to October.

"This is one of the best operational weather forecast models and is routinely used for forecasting hurricanes," he says. "It predicted the track and development of Hurricane Sandy very well, which is exemplary of the storms that might hit Europe in the future."

If Dr Haarsma is right, then the climate mechanisms driving hurricane-force winds are clearly changing. Indeed, climate scientists worldwide are publishing paper after paper trying to link such mechanisms to extreme weather.

"We have multiple lines of evidence indicating that climate extremes will increase," says Miguel Mahecha from Max Plack Institute for Biogeochemistry. "But the experiments that have been running for years have focused on increases in temperature, what I could call slow changes. We've never put an emphasis on short, abrupt events."

Mahecha is part of an EU-funded €3.3m collaboration – Carbo-Extreme – to integrate experimental data and predict the effects of climate variability and extreme events on the terrestrial carbon cycle. For example, land plants remove carbon dioxide from the air during photosynthesis, creating a huge carbon sink; but extreme weather events could change this.

"All evidence suggests a great number of extremes weakens the carbon sequestration potential... providing additional feedback to the carbon dioxide problem," says Mahecha. "Data from forestry bases and satellite records are being used to improve models and we are quanitfying gross impacts on the future. These models are the only way to measure future impact."

The UK's national weather service, the Met Office, is trying to link severe events to global warming. "It's not unreasonable to question whether we are seeing the effects of [manmade] climate change," says Peter Stott, head of climate monitoring. "Over the last decade, attribution science has progressed to the point that we can make 'attribution statements' about individual events."

As Stott highlights, such studies indicate the European heatwave of 2003 was 'very likely' to have resulted from manmade global warming. However, while the 2011 Thailand floods were extreme, the precipitation was not, and factors such as changes in river management more likely exacerbated effects.

A group of researchers formed the Attribution of Climate-Related Events group to build a system based on climate model simulations, to attribute extreme weather events to natural or manmade factors. The project has just received further EU funding to develop a 'quasi-operational' system over the next two years, which will only add weight to researchers' climate models.

Arctic research

In the aftermath of Hurricane Sandy, many researchers suggested reduced Arctic sea ice, due to global warming, altered the position of the jet stream, changing the track of the hurricane. Today's attribution systems cannot quantify this; but in two years this could change.

Attribution or not, some researchers don't need to wait. Professor Douglas Clark from the School of Environment and Sustainability, University of Saskatchewan, has been working in the far north of Canada for some 21 years. He says: "The magnitude of change is sobering."

Highlighting data from US National Snow Ice and Data Center that indicates an all-time low for the extent of Arctic ice in 2012, Prof Clark also says ice volumes are down 75 per cent in the last three decades, thanks to thicker, multi-year ice almost disappearing. It's no secret that temperatures are warming faster in the Arctic than the rest of the world; as sea ice melts it gives way to a darker ocean that absorbs more incoming solar radiation, amplifying the warming trend.

Prof Clark believes the effects will be profound. "These changes will continue to interact in ways we can't predict," he says. "What happens doesn't stay in the Arctic, it has implications for the global climate."

Unpredictable weather tracking

In a recent comment in Nature, Hajo Eicken from the University of Alaska, Fairbanks, stated that Arctic sea ice needs better forecasts to reduce hazards in its fast-changing waters. Most data used in forecasts are derived from thick, old ice, which dominated the ice pack until a few years ago. A lot less is known about the physical processes behind today's mix of young and old ice.

His sentiments echo previous research findings that show, for example, satellites are no longer keeping track of Arctic ice. And so, like their counterparts further south, Arctic scientists are now looking to share observations and data to improve forecasts and drive model development forward.

But as Prof Clark cautions: "The Arctic is unpredictable. Changes of considerable magnitude, the kind that were predicted back in the first Earth Day of 1973, have now been underway for some years. We are now very clear that we don't know exactly where we are going."