Photo: AFP
Photo: AFP

The margins for a drought

On an exam, you’d probably be happy with 86% or 88%; with the monsoon, those numbers spell bad news indeed

It must say something that when we hear predictions about the next monsoon, the stock market responds. As of today, it’ll be well over a month before the skies will bucket down on us. We always hope it will be a bountiful monsoon, of course. But in recent days, weather experts have actually predicted a good monsoon. That prompted a boom in market indices.

This cause-and-effect tale speaks of two large stories: one, the scale of the drought two years of poor rains have caused; and two, how dependent India remains on the monsoon. Since folks more knowledgeable than me have written plenty on those themes, I’d like to examine this idea of a normal monsoon, and the predictions around it.

First, how much rain do we get in India on average? By the simple device of monitoring rainfall data over many decades, the India Meteorological Department (IMD) calculates that India’s average annual rainfall (what it calls the “long period average" or LPA) is 89cm. What does that mean? The IMD explains that it is “the depth to which (the rainfall) would cover a horizontal projection of the earth’s surface, if there is no loss by evaporation, run-off or infiltration".

That is, it’s as if the monsoon left all of India under 89cm, or nearly 3ft, of water. That’s a lot of water. Remember, though, that this masks rainfall variation across the country. Some areas get so much that they flood; others get so little that they are virtually deserts. (This variation happened in 2015 too, despite the drought.) Famously, Cherrapunji is drenched in about 1,000cm of rain every year. In contrast, Jaipur gets about 60cm; points further west in Rajasthan get even less.

The IMD defines a “normal" monsoon as one in which we get between 96% and 104% of the average—or between 85.5cm and 92.5cm of rain. Between 90% and 96% of the average is considered “below normal", and anything below 90% is “deficient".

Now consider this: After the disappointment of the 2014 monsoon, which gave us only 88% of the average, the 2015 monsoon brought us just 76cm of rain, or 86% of the average. In an exam, you’d probably be happy with 86% or 88%; with the monsoon, those numbers spell bad news indeed. Such are the margins, when it comes to drought.

This is why so much of India—perhaps a quarter of all Indians, the government told the Supreme Court this week—is staring at drought. This is why so many of us are hoping so desperately for a better monsoon this year.

With the IMD’s latest forecast, that hope has just escalated. In its own language: “Quantitatively, the monsoon seasonal rainfall is likely to be 106% of the LPA." That is, the IMD thinks we’ll have an “above normal" monsoon this year, with just over 94cm of rain. Just 18cm separates last year’s deficient rainfall from this year’s forecast. That’s all it took to set the stock markets aflutter.

Such are the margins, indeed.

Still, it’s worth remembering that this is, after all, a forecast. Forecasting weather is a famously difficult business, even if it is now based on years of figures and ever-more sophisticated statistical models. This is why the IMD’s prediction comes with words like “likely" and a table of probabilities.

So, what their forecast actually says is, there’s a 30% chance that the monsoon will be normal, a 34% chance that it will be “above normal", and a 30% chance that it will be “excessive". (Much smaller residual probabilities for “below normal" and “deficient".)

Forecasting, as you can tell, remains an inexact science.

If you read their various bulletins carefully, the IMD acknowledges as much. “The success rate of IMD forecasts since 1988 has been high," they say on one of their webpages. In 19 of the 21 years from 1988 to 2008, their “forecasts were qualitatively correct… the exception was during 2002 and 2004, both of which were drought years".

But think what that phrase “qualitatively correct" also means: in the years 1994, 1997, 1999, 2002, 2004 and 2007, this IMD bulletin continues, “the forecast error—difference between actual rainfall and forecast rainfall—was more than 10%". What’s more, in 2012, 2013, 2014 and 2015, actual rainfall was less than IMD’s April forecast.

What if that happens in 2016 too, or what if we have a 10% forecast error this year? We could end up with a lot more rain than predicted. Given how bad the drought is, that would be welcome despite the threat of floods.

But equally, we could end up with 84cm or less of rain. That, according to the IMD, would be “below normal".

Knowing a little about what’s involved, I have enormous respect for experts who forecast the weather. Still, before feeling optimistic about an end to drought, I’d like to see the rain actually bucketing down.

Once a computer scientist, Dilip D’Souza now lives in Mumbai and writes for his dinners. A Matter of Numbers explores the joy of mathematics, with occasional forays into other sciences.

Comments are welcome at To read Dilip D’Souza’s previous columns, go to

My Reads Logout