This Indian Start-Up Might Predict This Year's Monsoon Better Than The India Meteorological Department!

This year will be unforgettable for Jatin Singh, CEO of Delhi-based Skymet. With the gumption that is characteristic of startups, Skymet is a young weather forecasting agency that has, with gradually amplifying audacity, been challenging the monopoly of the India Meteorological Department (IMD), the hoary state-led colonial-era institution, on matters related to climate and weather in India.

The arena where these two unequally balanced gladiators face off is no festive cricket pitch but in lumbering office blocks stacked with behemoth supercomputers and glitzy large flatscreens that are constantly monitoring the grey, gloomy sky. Scientists employed by both agencies scour, measure, calculate, theorise and ultimately guess when, where and how much rain-water these temperamental skies are going to disgorge over India's rain-dependent fields.

The IMD has usually been the final word on droughts but this time Skymet has asserted that the IMD is grievously mistaken. The IMD expects India to get 'below normal' rains during the coming monsoon months between June and September whereas Skymet says farmers and citizens needn't be worried: India is going to have 'normal' rains.

This is glaringly incongruous because the IMD doesn't like rocking the boat. Between 2002 and 2013, when India saw four drought years, the IMD never gave any indication in its April forecast that the country should be preparing for a drought. It was only last year that IMD, for the first time, said in April that rains would be 'below normal', a prediction that eventually proved right. This year it's again playing harbinger of bad news and it is the up-start private agency that seems to be trying to infuse optimism by declaring normal, well-distributed rainfall.

"It is too early to say things like drought but we're fairly confident that El Nino conditions [a weather phenomenon with negative consequences for the monsoon rains] will prevail and there don't seem to be other factors yet that will strengthen the monsoon," said DS Pai, chief forecaster, IMD.

IMD is still India's only official forecast agency. Though other research organisations (all of them either universities or funded by the Centre) also work on monsoon models, the IMD usually 'incorporates' and averages out their differing predictions into its own middle-of-theroad 'safe' output.

et

The Forecast Game

Though Skymet has been formally issuing forecasts only since three years, compared with the IMD's more than 100, their predictions during this period haven't been widely apart. Last year, both organisations said India would get deficient rains and the country did end up with what is called a 'meteorological' drought.

In 2013, both agencies rightly predicted that India would get bountiful rains though Skymet was closer to the mark. "This time we've really stuck our neck out," said Singh and though he follows everything that the IMD does extremely closely, he finds no reason to alter his prediction ¡ª made a few weeks before IMD's ¡ª that India will get good rains.

The monsoon rains between June and September, which irrigate threefifth of the sown area, are critical for Indian agriculture. Though over the years India has managed to build up sufficient buffer stocks of grain and ensure that ¡ª poor rains notwithstanding ¡ª most of its poor can buy essential rice and wheat, the performance of the monsoon rainfall has a direct bearing on the robustness of the rural economy.

If farmers are told that rains are likely to be weak, they'd rather leave their farms fallow and spend less on a variety of commodities ¡ª from fertilisers to refrigerators ¡ª that season, all of which incrementally adds up and depresses the economy. Weak rains can also force the government to import crops such as pulses, oilseed and sugar and thereby add to inflation.

For preparing its forecast, the IMD relies on complex statistical relations based on troves of weather data that it has gathered since the 19th century. Forecasters correlate the likelihood of weak or abundant rain with a handful of selected climate variables such as sea surface temperatures in the Central Pacific or snowfall in Europe.

et

Gruelling Method

Hundreds of years of such observations have helped IMD scientists identify diverse phenomena, including El Nino and La Nina, that affect rainfall patterns over the country. Between March and May, IMD gets values of certain climate phenomena based on which it issues two-staged forecasts, the first in April and then in June.

However, 2002 and 2004 saw some of India's most significant droughts and nothing in IMD's weather models indicated any sign that such rainfall deficits were imminent. In 2009, even after junking the mathematical techniques it had used for decades and adopting new climate variables, the IMD failed to anticipate a drought, India's biggest and most wide-spread in 40 years.

Moreover with climate change having firmly established itself as a truism and the sophisticated weather models crunched in supercomputers, the tools of choice to parse the weather, firms such as Skymet ¡ª that rely on methods such as these ¡ª have started to gain importance.

Fluctuating Weather

What differentiates the approach of both organisations this year is the likelihood of an El Nino, the anomalous heating of water in the central and eastern Pacific that frequently dries up rain in northwest and central India. Between 1880 and 2015, El Nino years have corresponded more frequently to drought than normal rainfall.

The IMD believes this year will be an El Nino and Singh says that is unlikely because the El Nino happened last year, which turned out to be a drought. "That's where the IMD and I differ. The El Nino has been present all of last year and will reduce in intensity by the time the monsoon months kick in," said Singh. He's also hedged his bets saying that even if 2015 were to become an El Nino year the chances of two drought years back to back were only a minuscule 3%. "So even statistically I don't think there should be worries about weak rains this year," he added.

Singh is in fact more worried about the searing rains that he says will continually lash India for much of May ¡ª typically supposed to be a dry-rain starved month ¡ª and contribute to instances of crop loss. Concerns over farmer distress because of unseasonal rains have rocked parliament and therefore rain ¡ª deficient or in excess ¡ª is likely to be a hot topic this year.

Though it isn't official, the IMD is crafting burial plans for its age-old statistical models. The Monsoon Mission, which is the poster boy for IMD's plans to adopt contemporary approaches ¡ª called dynamic forecasting ¡ª is a Rs 400-crore programme housed at the Indian Institute of Tropical Meteorology (IITM) in Pune.

The aim is to rely on powerful supercomputers that simulate the weather on any given day and then, based on equations of physics, extrapolate weeks and months into the future. It isn't that IMD has not thought on these lines before but that such a method, according to its scientists, is reliable for predicting the weather few days or weeks ahead, but fails quite badly when trying to look at the monsoon that spans four months.

"There is no single dynamical model that can work," said Madhavan Rajeevan, director of IITM. "We have to spend a few more years adapting several global dynamical models to Indian conditions. The aim is not be right once or twice, but to be consistent. I'm confident nobody in the country can yet be that."

Originally published in the Economic Times