Later this week, R.C. Bhatia, veteran meteorologist with the India Meteorological Department (IMD), will make this year’s monsoon forecast. Ahead of this annual ritual, the IMD chief spoke to Mint, explaining the nitty-gritty of weather forecasting models and the difficulty of predicting the Indian monsoon. Excerpts:
For four years, we’ve been getting the forecast in April. How far away are we from D-day?
We have not decided on the date yet, but hopefully, we should be out with it before next Saturday (21 April). We’re ready with the results from our models, and have received inputs from other research organizations that also prepare their own forecast models. All we have to do is compare results and take decisions.
So, it’s not the IMD alone which is involved in weather forecasting?
Officially, only IMD can issue forecasts, but we look at results from the Indian Institute of Tropical Meteorology, Pune; Space Applications Centre, Ahmedabad; Centre for Mathematical Modelling and Computer Simulation, Bangalore; and National Aerospace Laboratories, also in Bangalore.
Our main concern is drought, which is almost impossible to predict. The Bangalore models are dynamic and the others, statistical in nature. Now, if some models predict a drought, and the others normal rainfall, then we recheck our data, and maybe re-evaluate it. Eventually, we do reach a consensus on what our forecast should be and then hope for the best. On the day of the forecast, we will also make public the forecast by each organization.
Since you launched these new models four years ago, you’ve got your forecasts correct in alternate years beginning 2003. How confident are you of your models?
If IMD is using some models, it means they are the best and the most reliable of the current lot. See, the statistical models depend heavily on how accurately certain parameters are chosen. Wrong parameters mean deviant results.
Let me give you an example. From 1998 to 2002, we used a 16-parameter statistical model, which worked well until it went wrong on the 2002 drought. We later realized that some parameters in the model showed increasingly weakening correlations. The new eight- and 10-parameter models that we now use employ newer parameters, which we believe are better indicators of our monsoon.
All over the world, meteorologists use dynamical monsoon prediction models. Why are we still using statistical models?
Two main reasons. One, the Indian monsoon is such that dynamical models are simply unable to accurately model them. We have an extremely complicated climate, largely because of our range of geographical features, and so for now, we have to depend on the statistical models.
Secondly, our dynamical models are still in the experimental stage and need a lot more investigating and research.
Ultimately, the future lies in dynamical models as they will be able to give more useful data such as how the rain will spread out, which areas will receive more or less rain. We do use dynamical models for short-range prediction, such as for a day or week, but being able to predict the summer monsoon is beyond our abilities at present—at least over a decade away.
Since last year, IMD has been revamping itself and investing in hi-tech instruments. How will this help with our monsoon forecasts?
Immensely. We have installed about 75 automatic weather stations across the country. These help us gather data on rain, temperature, wind-speeds nearly four or five times a day. The rain-gauges currently installed provide data only once a day, and many a time that too is inaccurate.
More data means better understanding of how all these weather parameters affect each other. Better understanding means better equations to model the weather.
Ultimately, dynamical weather models are all about getting the right equations.
Will this year’s forecast use the same models as last year?
Summer temperatures this year are one-two degrees higher than last year. There were extended showers in March. Could they affect the monsoon?
Not really. Because all these factors are largely due to the Western disturbances (fluctuating temperatures over Europe and Central Asia). There’s no direct correlation between the Western disturbances and our rains, but there are signs of an emerging La Nina. Historically, a strong La Nina means no drought for India, but this can be confirmed only around June.
Graphic: Will it rain or pour?