When a 215 km/h Cyclone Hit Odisha, Forecasts Had Already Given India Time to Move a Million People

Contents

The Human StoryThe Technology Enters the StoryHow the Technology WorksThe Moment Everything ChangedWhat Would Have Happened Without the Technology?The Technology Behind the BreakthroughWhere the Technology Is TodayThe LimitsWhat Comes NextConclusionSourcesEditorial Disclaimer

Late on the night of 12 October 2013, a cyclone with sustained winds of about 215 kilometres an hour came ashore near Gopalpur, on the coast of Odisha in eastern India. India’s weather department called it the most intense cyclone to cross the Indian coast since the Odisha Super Cyclone of 1999, a storm on the same coastline that killed more than 10,000 people.

This time, the official death count ran into the dozens.

The two storms were not identical, and the comparison needs care. But one difference is not in dispute: in 2013, people had days of warning that were specific enough to act on. Close to a million of them had already left the coast by the time the eye of the cyclone arrived.

This is the story of what that warning was made of, who it reached, and what it could not do.

The Human Story

On 29 October 1999, a super cyclone struck the same stretch of Odisha. The World Bank’s later account puts the death toll at more than 10,000 and describes massive destruction of housing and public infrastructure. An account published by Al Jazeera in 2013 says that in 1999 just 44,500 people had been housed in 23 evacuation shelters.

After that disaster, Odisha set up the Odisha State Disaster Management Authority, which the World Bank describes as the first state agency in India devoted entirely to disaster management. Over the following years the state kept building cyclone shelters and evacuation routes, strengthened coastal embankments, and ran mock drills every year in its coastal districts. It also worked with communities to set up volunteer teams who knew what to do when a warning came.

Then came October 2013. The storm began as a remnant circulation from the South China Sea and became a depression over the Andaman Sea on 8 October. By then Odisha’s officials were already working from forecasts that said where it was likely to go.

A note on the human side of this story: the sources I could verify do not name individual evacuees, so I will not invent a family or a village. The people in this story are best described the way the records describe them, as roughly a million men, women and children who moved to shelters, safe buildings and higher ground before the storm arrived. Odisha’s chief minister, Naveen Patnaik, told television viewers that almost nine lakh (900,000) people had been evacuated, according to an NPR report.

The counts vary by source, which is normal for evacuations of this size. A World Bank summary gives around 850,000 in Odisha and 150,000 in Andhra Pradesh. The weather department’s report, quoting state figures, lists 983,642 people evacuated in Odisha because of the cyclone and 134,426 in Andhra Pradesh. The honest summary is close to a million, and in the largest pre-disaster evacuation in India’s history, according to Indian officials and the reporters who covered it.

The Technology Enters the Story

A cyclone like Phailin is watched long before it arrives, and by several kinds of instrument at once.

India’s weather department reports that it tracked the storm using satellite pictures taken every half hour from its INSAT and Kalpana weather satellites, supported by microwave imagery, satellite wind measurements over the ocean, buoys, coastal weather stations and automatic weather stations. On 12 October, when the cyclone came within range, a Doppler weather radar at Visakhapatnam began watching it continuously from 6:30 in the morning, when the storm was about 310 kilometres offshore, producing a new image every ten minutes.

Behind all of that sat computer models. The department runs its own regional and global models and compared them with those of the UK Met Office, the US National Centers for Environmental Prediction, the Japan Meteorological Agency and Météo-France.

The result was a timeline that gave officials something concrete to plan around. On 8 October, when the depression formed, the department forecast that it would become a cyclonic storm and move toward the coast of north Andhra Pradesh and Odisha. On the morning of 9 October, more than three days before landfall, its bulletin said the storm would become a very severe cyclone and cross the coast between Kalingapatnam and Paradip on the night of 12 October. The forecast track pointed near Gopalpur, which is where the storm came ashore.

How the Technology Works

Think of forecasting a cyclone as three jobs: seeing it, simulating it, and telling people.

Seeing it. Weather satellites watch from space. Among other things they measure how cold the tops of clouds are: tall, towering storm clouds are far colder than the sky around them. The department’s report notes that on 8 October, as the system was forming, the coldest cloud tops were about minus 70 degrees Celsius. Radar does something different. It sends out pulses and listens for the echo from raindrops, which lets forecasters see the shape of the storm, including the calm eye at its centre, and track how it moves.

Simulating it. A weather model divides the atmosphere into a grid of boxes and uses physics to calculate how air, heat and moisture will move from one box to the next, over and over. The department’s regional model, for example, runs at grid sizes of 27, 9 and 3 kilometres. No single model is always right, so forecasters compare several and look for agreement. The department’s report says the models disagreed about the landfall point in the early days, but that their answers converged as the storm came closer, and that its own combination of models, which averages several together, gave the best landfall forecast.

Telling people. The cone of uncertainty, the widening shape drawn around a forecast track to show where the centre might go, was published for the storm. The state’s cyclone warning centre in Bhubaneswar issued 21 cyclone warnings before landfall and 38 bulletins for the press, All India Radio, Doordarshan and other TV channels, according to the department’s report. The World Bank notes that the project it was financing in Odisha and Andhra Pradesh focused on early warnings that reach down to the last community, backed by shelters, evacuation roads and stronger embankments.

The Moment Everything Changed

There was no single dramatic moment, which is the point. The moment was a decision, made early: act on the forecast.

The storm made that decision harder in one way. Between the morning of 10 October and the morning of 11 October, the cyclone strengthened by about 70 knots, according to the department’s report, going from about 45 knots to 115 knots (around 215 kilometres an hour). On 11 October the department raised its forecast wind speed at landfall to 210 to 220 kilometres an hour, and the storm crossed the coast with sustained winds of 200 to 210 kilometres an hour, gusting to about 220.

The landfall forecasts, by the department’s own measurement, were remarkably close. It reports that the error in its forecast landfall point was between 3 and 13 kilometres for forecasts made 12 to 72 hours ahead, smaller than the 15 to 20 kilometre eye of the storm. The timing was off by between one and three hours. Its average error for similar forecasts over the previous years was far larger. Those are the department’s own figures, and a verification report written by the forecaster itself deserves some caution, but they match what the evacuation records show: the right places were emptied, in time.

What Would Have Happened Without the Technology?

Nobody can know exactly. What the records support is narrower than “technology saved a million lives.”

In 1999, according to Al Jazeera’s account, just 44,500 people were in shelters. In 2013, the warning came days ahead and named a stretch of coast, and the state moved close to a million people. The official toll stayed far lower: a World Bank summary recorded 38 deaths 72 hours after landfall, and the weather department’s report lists 21 deaths from the cyclone and 17 from flooding in Odisha, plus one in Andhra Pradesh. Other tallies run higher when later flooding is counted, so the exact figure depends on what is included, but every version is in the dozens, not the thousands.

The comparison with 1999 is not like for like, because the storms differed in strength, surge, landfall location and how many people live on the coast. And the technology was not acting alone. Odisha’s disaster agency, shelters, embankments, volunteers and yearly drills were, in the World Bank’s account, what turned a forecast into a survived night. A forecast that nobody acts on saves no one.

The fairest way to put it is that the technology bought time, and years of preparation spent that time well.

The Technology Behind the Breakthrough

The unusual part of Phailin’s forecast was how well several systems worked together.

Satellites supplied the early view and the cloud-top measurements. Radar took over once the storm came close. Models, run at different resolutions and compared against each other, produced the track. A digital system the department calls the Tropical Cyclone Module pulled the guidance together so forecasters could compare it quickly and produce warnings. Separate storm-surge forecasts used models from India’s ocean information service and IIT Delhi. After landfall, a survey found the surge was about 2 to 2.5 metres above the normal tide in Ganjam district, with flooding reaching up to a kilometre inland in low-lying areas.

The department’s report also shows how forecasts had improved. Its average track error for 72-hour forecasts of Phailin was about 77 kilometres, compared with a 2008 to 2012 average of about 376 kilometres. That one comparison cannot prove a trend, because one storm is one storm, but it shows the scale of the gap the forecasters were reporting.

Where the Technology Is Today

This article describes the systems that existed in 2013, as documented in the department’s report and the World Bank’s account. I did not find a verified, current description of India’s cyclone forecasting for this article, so I will not guess at today’s setup.

What the 2013 sources do say is that the system was still growing. The department notes that five-day (120-hour) forecasts were introduced in 2013, the year of Phailin. And the World Bank was funding a national cyclone risk project, running since March 2011 in Odisha and Andhra Pradesh, aimed at improving the last stage of warning, the part that reaches individual villages.

The Limits

A good story about forecasting should be honest about where forecasts fall short.

• Intensity is harder than track. The department’s intensity forecast errors in 2013 were higher than its 2008 to 2012 average. Its report attributes this to the rapid strengthening on 10 October. Its early forecasts were also lower than what happened: the 9 October bulletin expected winds of 175 to 185 kilometres an hour at landfall, and the storm arrived with 200 to 210.
• Forecasters disagreed. Foreign agencies forecast a stronger storm than India’s weather department did. Reporters at the time noted that US forecasters had warned of a “super cyclone,” while the department said the actual storm was a very severe cyclone, comparable to a Category 4. Al Jazeera also noted that wind estimates from satellites carry considerable margins of error, and that India’s weather department has no aircraft for direct measurements.
• Damage was still severe. The department’s report lists about 12.4 million people affected in Odisha, 419,052 houses damaged and 668,268 hectares of crops affected. Odisha’s disaster management minister put crop losses above 500,000 hectares, worth about 24 billion rupees. A mild death toll does not mean a mild storm.
• Communications can fail. Mobile networks and power were still being restored after landfall, according to CS Monitor. A warning system that depends on phones needs a back-up for the hours after the storm.
• The numbers are not settled. Death tolls and evacuation counts differ between sources, as described above.

What Comes Next

Already in use: multi-model forecasting, radar tracking close to the coast, published cones of uncertainty, five-day forecasts, and shelter and evacuation programmes of the kind Odisha built.

Suggested by the 2013 data: the weak point shown by Phailin’s numbers was intensity, especially when a storm strengthens quickly. That is my interpretation of the department’s own error tables, not a prediction from a source.

Not claimed here: this article makes no predictions about future technology that I could not verify.

Conclusion

The most striking number in this story is not 215 kilometres an hour or a million evacuees. It is the gap between the warning and the storm: more than three days between a forecast that named a stretch of coast and a cyclone that arrived there, within a few kilometres of the prediction.

A satellite and a radar can only report what the sky is doing. It takes a state that built shelters before it needed them, volunteers who knew the plan, and officials who trusted the forecast enough to move a million people on the word of a weather map. But the story begins with the forecast, because without those days there would have been nothing for the shelters and drills to work with. Here, technology did not stop a cyclone. It gave an entire coastline the time to get out of its way.

Sources

Source	Type	Published	What it supports
Very Severe Cyclonic Storm, PHAILIN over the Bay of Bengal (08-14 October 2013): A Report, India Meteorological Department, Cyclone Warning Division	Government agency report (primary source; 43 pages, read in full)	October 2013	Storm life history, landfall time and wind speed, rapid intensification, observation tools (satellite, radar, buoys), models used, forecast timeline, landfall/track/intensity forecast errors, bulletins issued, damage and evacuation figures, storm surge survey
Cyclone Devastation Averted: India Weathers Phailin, GFDRR (copy of the World Bank story)	Development institution (World Bank / GFDRR)	2013 story, page dated 24 May 2014	1999 toll (10,000+), 38 deaths at 72 hours, approx. 850,000 + 150,000 evacuated, OSDMA, shelters, drills, NCRMP project details
Speedy Evacuation In India Saves Lives During Cyclone, NPR (transcript)	Established news organization	14 October 2013	Chief Minister Naveen Patnaik’s statement of almost 900,000 evacuated; 1999 toll
In India, Cyclone Phailin’s devastation tempered by planning, The Christian Science Monitor	Established news organization	13 October 2013	Evacuation scale, forecaster disagreement (Category 5 warnings vs. IMD), crop damage figure, network restoration
Evading the wrath of Phailin, Al Jazeera	Established news organization	13 October 2013	1999 shelter comparison (44,500 people in 23 shelters), more than 500,000 in shelters in 2013, forecast disagreement, wind-estimate uncertainty

Two further sources, an Indian Journal of Forensic and Community Medicine review and an ICMR case study on Phailin, were only partly seen and are used only to flag that death-toll counts vary.

Editorial Disclaimer

This article is an independent editorial work for Tech Horizon City’s “Technology That Changed Lives” series, based on the publicly available sources listed above. Figures such as death tolls, evacuation counts and forecast errors differ between sources and are reported here with their source. The forecast-accuracy figures come from the India Meteorological Department’s own report. If any factual error is identified, Tech Horizon City will correct it promptly after verification.