Never know where it will hit till it hits, says Nookaraju, a philosophical farmer in Andhra Pradesh state, along India's southeast coast.
Millions of farmers along the east coast, with its considerable river systems, low-lying deltas, and scattered islands, dread the path of a furious cyclone and deeply empathise with the millions in Myanmar (formerly Burma) who lost their lives, families and property when cyclone Nargis struck in May.
Last year, it was Bangladesh's bad luck [when cyclone Sidr struck in November 2007]. For all you know it will be our [India's] turn next. Bengal, Orissa, Andhra or Tamil Nadu who knows?, says Nookaraju.
Yet India's meteorological scientists hope to soon have some answers. India is at the forefront of South Asian efforts to improve cyclone prediction, and hopefully give a life-saving chance to locals like Nookaraju.
Bay of Bengal storms
Cyclones known as 'hurricanes' in the North Atlantic and eastern Pacific Ocean, and 'typhoons' in the west Pacific Ocean are low pressure wind systems that build up over tropical and sub-tropical waters.
Spiralling cyclone winds churn up ocean waters. Giant waves and storm surges rush inland, flooding low-lying coastal areas.
Uma Charan Mohanty, professor at the Centre for Atmospheric Sciences at the Indian Institute of Technology in Delhi, says there are 80100 cyclone events each year worldwide.
Only three or four cyclones affect Asia's Bay of Bengal each year, yet they can cause more damage than the faster and wider reaching typhoons and hurricanes elsewhere in the world. The shallow, funnel shape of the Bay of Bengal helps these cyclones churn up waters fast, throwing up 810 metre high walls of water that strike the coast.
Overall, the Indian sub-continent generates only seven per cent of the globe's tropical cyclones, but suffers the highest number of human deaths from them, says Mohanty, who also served at the Asia Pacific Centre for Disaster Preparedness in Bangkok until May 2008.
In 2005, 25 hurricanes in the Atlantic caused ten deaths. In contrast, cyclone Sidr in 2007 left 3,500 dead.
Part of the reason is the disparity in warning systems and evacuation measures. Many countries bordering the western Atlantic have good warning systems that give ample time for local people to prepare or move to a safer area. The populous and poor Asian coasts struggle with poor warning systems that give less advance notice. Rains and strong winds lash areas in the cyclone's path a day ahead of it, hampering evacuation. And the Bay of Bengal cyclones build up over four or five days, unlike their Atlantic counterparts that take eight to ten days the short duration making it more difficult to prepare and take emergency measures.
Even with an upgraded warning system, most of the poor coastal farming and fishing communities in Asia would still find it difficult to evacuate. They have little private transport and live in areas with poor to non-existent roads.
We need an accurate warning system of at least 4872 hours, says Mohanty.
He adds that another problem is that false warnings from an inaccurate system may desensitise coastal communities. In Bangladesh, for example, some communities, fed up with a false tsunami warning after an Indonesian earthquake in September 2007, ignored the cyclone Sidr warning issued a month later.
Poor communities are expected to become more vulnerable, with scientists predicting that global warming will lead to fewer but stronger cyclones in ocean basins worldwide (see Global warming 'induces fewer, but meaner, cyclones').
Scientists at the Indian Institute of Tropical Meteorology (IITM), in Pune, are investigating how global warming is likely to affect cyclones forming in the northern Indian Ocean.
IITM uses PRECIS, a high-resolution regional climate model developed by the UK's Hadley Centre for Climate Prediction and Research, to simulate climate change up to 20712100.
The simulations study how climate change might affect the frequency and intensity of 'cyclonic disturbances', which range from severe storms and gales to low-pressure areas that may not form a storm.
IITM's studies indicate that while there may be no significant change in the frequency of cyclonic disturbances under different warming scenarios, their intensity seems likely to increase by ten per cent by 2100, compared to present day events.
Kanikicharla Krishna Kumar, who heads the climate change research centre at IITM, says these are extremely preliminary results, based on one simulation from the regional climate model, but may be taken as one of the possible future scenarios.
If cyclones are to become more dangerous, prediction will need to become more accurate.
Cyclone prediction combines predictions of ocean and land weather systems. This includes detecting and tracking cloud clusters from which damaging winds and rain can develop, how these will develop into a storm, how a cyclone core is developing and strengthening, the track (path) of a cyclone as it approaches land, and storm surges as the cyclone moves inland.
Scientists are striving to predict cyclone timing, location and intensity more accurately.
The current 48 hour forecast in the Bay of Bengal can predict [over] a 250300 kilometre area. This is too large for evacuation. We need to narrow down to at least 100 kilometre accuracy, says Dev Raj Sikka, former head of India's climate research programme under the Department of Science and Technology in Delhi.
Improved forecasts require both good models and good-quality data. India has to work on both.
Scientists generally rely on atmospheric global circulation models (GCMs) for cyclone forecasting. But these models are too coarse to provide a realistic description of smaller scale events, such as many tropical cyclones. To improve the accuracy of cyclone forecasts, GCMs need to be 'downscaled' to a resolution of better than ten kilometres. Indian meteorologists are currently training at the National Centre for Atmospheric Research, Boulder, USA, to do this.
And while predicting cyclone track has improved over the past three decades, there has been no significant improvement in forecasting their intensity. Current statistical prediction models can predict the general trend in intensification but cannot measure the actual rapid changes in intensity.
Scientists also need to improve how they predict cyclone frequency and intensity over long time scales. And they need to link their forecasts of where a cyclone will hit with the surrounding topography whether the land is rocky or flat, or the surrounding sea deep or shallow, for example to determine the impact.
An eye on the storm
India is moving towards a much improved cyclone monitoring system.
The country has a network of ocean buoys to gather data on sea surface winds, pressure and temperature.
It also uses a constellation of satellites to gather weather and cloud data and provide hourly information on how cyclonic systems are evolving and intensifying.
Five Doppler radars along the east coast detect storms building up within 300400 kilometres. Coastal radiosondes (probes for taking atmospheric measurements) are released to pick up winds every six hours, and three ocean research ships regularly release balloons with sensors.
The Indian Meteorological Department (IMD), the World Meteorological Organization's specialised centre for Asia, plans to further upgrade and automate its observation systems. It will add 12 more Doppler radars over the next three years, and boost the number of land-based automatic weather observation systems to 550 during 200809, and to 1,000 by 2011. There are also plans for 1,000 new automated rain gauge stations during 200809, and 4,000 over a longer period, says Rajendra Hatwar, the IMD's additional director general of research.
But as Sikka points out, India cannot yet collect data from the central 'eye' of the storm, and that's what dictates a cyclone's movement.
India uses data on winds outside the core region to estimate parameters within it. Unless India improves its data, considerable improvement in the accuracy of forecast models cannot be expected, says Sikka.
For this, aircraft reconnaissance is needed. A helicopter penetrates the core of the cyclone to drop a balloon equipped with sensors to measure wind speed, pressure direction and humidity.
The United States is the only country to have aircraft reconnaissance which helped improve [cyclone] track prediction in the Atlantic to a 150 kilometre range over the past 20 years. In India, track prediction has remained stagnant at 300 kilometres, says Sikka.
At an Indo-US workshop on tropical storm forecasting, held in Delhi in March, India and the United States agreed to work together to generate additional data from the Bay of Bengal during 200809. Under this pilot project, the US will provide expertise and India will use its own aircraft to gather data from the eye of the cyclone (see India to employ aircraft in cyclone forecasting).
And by 2011, the IMD plans to have a fully operational aircraft probe, with US assistance.
Indian scientists are also striving to reduce the error in forecasts of cyclone track by developing improved models. At present this may be inaccurate by as much as 140 kilometres for a 24 hour forecast, 240 kilometres in a 48 hour forecast and 310 kilometres in a 72 hour one.
All these measures should help predict where and when cyclones will hit. But the major challenge remains to translate these advances into timely action that can save lives. Only then will there be real progress in the battle against cyclones.
This article is part of a Spotlight on Tropical cyclones in the Indian Ocean.