Monsoon model from statistics to simulation

The Indian weather office will change its monsoon forecasting method yet again as it tries to call correctly the naturally occurring phenomenon with the greatest impact on economic growth. The Meteorological Department failed to predict droughts that hit the country in three years this decade, putting growth under pressure and causing inflation to rise. 

The new monsoon forecast model, likely to be in place by 2015 after work begins on it by the end of the year, is expected to do away with the current practice of monsoon forecasting based on statistical models. Instead, it will simulate meteorological conditions such as wind speeds and ocean temperatures to forecast rainfall. The effort will be to improve a model developed by the US National Oceanic and Atmospheric Administration, or NOAA.

Officials in the ministry of earth sciences said NOAA will provide this model, currently referred to as “NCEP-CFS” (National Centre for Environmental Planning-Coupled Forecast Model) to Pune-based Indian Institute of Tropical Meteorology (IITM). “Once we get that, we will organize a large group of leading weather scientists, hopefully about a hundred, and they will dedicatedly work on this model, improve it over time and ultimately introduce this as part of our operational (official) forecast system,” said Shailesh Nayak, secretary, ministry of earth sciences.

NCEP-CFS will be what meteorologists call a dynamical atmospheric-ocean coupled model. In this approach, a spectrum of meteorological data from wind speeds and sea surface temperatures is used to simulate the weather. Along with the numbers, equations that govern the relationship between these data are fed to supercomputers that crunch and extrapolate the forecast over required time frames.

Using such an approach to predicting climate means scientists don’t need to depend on tenuous historical data or statistical relationships between atmosphere phenomena, such as snowfall over Europe, and historical trends in monsoon rainfall. Statistical models are generally acknowledged to be unreliable by weather scientists as correlationships between variables could randomly fail any year. Yet, the IMD still relies on them as no dynamical model, Indian or foreign, can forecast Indian monsoon beyond a week.