130 years on, IMD says still trying to know monsoon

130 years is long enough to master the behaviour pattern of India’s lifeline, the south-west monsoon, or so one would think.

Not so, says the India Meteorological Department (IMD). It’s still trying to “understand” the phenomenon, let alone predict it.

“First is to understand monsoon better. Prediction is the next stage. We have to understand so many things about the monsoon,” director general meteorology Ajit Tyagi said. No wonder, India’s apex weather bureau, which made its first tentative monsoon forecast in 1882, finds it difficult to get its predictions right.

In 2011, seven out of 10 southwest monsoon forecasts were off the mark.  The overall rainfall in the four-month season ended September was 101% of the long period average, compared with the department’s prediction of 91-99%.

Tyagi says the sudden re-emergence of La Nina conditions during August-September is partly to be blamed for the predictions going awry this year.

La Nina, the cooling of sea surface temperatures in east central Pacific Ocean, is usually associated with increased rainfall in India.

“When we knew La Nina is taking place, we were sure we will be a little bit off (the mark),” Tyagi said. He also says the models that the department currently uses for long-range forecasts may not be best suited to pick up such unexpected developments.

“This is the problem with statistical forecasts. The dynamical forecasts may update the things (better),” Tyagi said. The department relies on statistical models for its forecasts, as the skills of dynamical models are yet to be proved. “The spread of dynamical models used to be very wide, they were never able to simulate monsoon rainfall very realistically.”

For instance, the monsoon forecasts on the dynamical models in June ranged between sub-normal to above normal rainfall, he said.

“There were (dynamical) models giving sub-normal rainfall. There were models, which were giving 114% also,” he said.

In June, the bureau went with the statistical model and lowered its monsoon rainfall forecast to 95% of the long period average from 98% estimated earlier. The actual rainfall in the season turned out to be 101%.

“Skill of these (dynamical models) still is not superior to statistical models. When final forecast is given, we still give heavy reliance to statistical models,” Tyagi said.

However, in view of the recent experience, the weather department plans to re-look its prediction models, he said.

“...There may be review of this particular incident also. (We will see) Whether we require some refinement in our statistical model. Also, whether we give over-reliance to statistical model or we should now start giving equal amount of confidence to the dynamical models also,” Tyagi said.

Even if the weather bureau refines its models, it may still not be able to predict extreme events like drought, Tyagi said.

“Globally, nobody can predict extreme events,” Tyagi, who headed the Indian Air Force’s meteorological branch before taking over as head of India Meteorological Department in 2008, said.

Even if the predictions throw up extreme events, the department will be wary of using it because of the uncertainties in the models, he said.

“Even if I have got some indications (about extreme events) with so much of uncertainties in the model I can give (only) a general direction,” Tyagi said.