India is building a new system to predict the monsoons - one that will forecast summer rain more accurately than the present model.
This comes after the ministry of earth sciences, which oversees the Met department, concluded that its 'empirical method' has become obsolete.
The new model - called the dynamic model - will cost Rs. 400 crore and take three years to be ready.
"Though the empirical model is still relevant for many countries, we can't use it effectively in India anymore," Shailesh Nayak, secretary, earth sciences ministry, says.
The June-September rains are a national fixation due to their impact on the economy. Two-thirds of Indians depend on income from farms, 60% of which do not have irrigation.
In the last 25 years, the Indian Meteorological Department has got its predictions wrong almost half the time. This year, too, the monsoon forecast had to be downgraded thrice.
"Our skills in monsoon forecast have improved, but not sufficiently," says Nayak.
Since 2007, monsoons are being predicted through an improved version of the empirical model, which has been around for over a century.
This year, Met scientists made experimental forecasts using the dynamic model.
"But it is not ready for official forecasts yet," Nayak says.
The dynamic model relies on the US's 'couple model', which uses atmospheric and oceanic data, and the UK's 'unified system model', a single forecast system for both long-term and medium-term forecasts.