India to Predict Monsoon Rains With Supercomputer Accuracy

Monsoon Rains Roll Over Mumbai

Enygmatic-Halycon via Wikimedia

In an effort to stay one step ahead of the summer monsoon season, Indian scientists are embarking on an ambitious and unprecedented project to build computer models that will allow them to predict the movements of erratic monsoons weeks in advance. If successful, the Indian government thinks it can drastically alter economic outcomes for hundreds of millions of people whose lives depend directly on India's agriculture sector.

The ability to predict the movements and impacts of monsoons ahead of time would go a long way toward augmenting India's disaster management services, but in the long term it's very much more important to the economic health of the nation, which derives 15 percent of its economic prosperity to agriculture. The June-September monsoon season provides three-quarters of India's annual rain during a "normal" season--though this summer's below-average rains have led to fears of drought and famine even though some parts of the country have seen flooding that has displaced two million people.

With these problems in mind, Indian computer scientists are working with American and British scientists to hack the monsoon season, tapping supercomputers to create both long- and short-term computer models that provide far more detailed and actionable information. This not only helps farmers tilling the land know how to better manage resources at field level, but also would help Indian authorities better conserve water resources and craft more-informed economic policies.

Right now, farmers get long-term projections twice a year--once in April with an update in June--detailing what the Indian Meteorological Office (IMD) sees on the horizon. These forecasts are based on wind speeds, sea temperatures, and air pressure compared with similar data over the past five decades. It's imprecise and, as weather patterns the world over are shifting, increasingly inaccurate.

The new monsoon mission wants to offer short-term forecasts accurate out to 15 days, and long-term seasonal forecasts that are better than the ones it issues now, based on more granular data and better computer modeling. That's a difficult computing challenge, but one that's not outside the realm of possibility. IBM's Deep Thunder project already does something like this for regions in New York, Brazil, and elsewhere. Where there is data and computing power, there is potential. India hopes to leverage that into increased economic stability for one of the world's most populous nations.