The annual climate summit opened in Cancun, Mexico, this week. A few days earlier, while releasing a new report, Climate Change and India: A sectoral and regional analysis for the 2030s, environment and forests minister Jairam Ramesh emphasised, “It is imperative” that India has “sound, evidence-based assessments on the impacts of climate change”. The report claims that India will soon be able to forecast the timing and intensity of future monsoons that are so critical to its agricultural base.

Could 250 of India’s top scientists be wrong when they say their computers will soon be able to predict summer monsoon rainfall during the 2030s, based on projected CO2 trends? Do scenarios generated by climate models really constitute “sound, evidence-based assessments”? We do not believe it is yet possible to forecast future monsoons, despite more than two centuries of scientific research, or the claims and efforts of these excellent scientists. The Indian summer monsoonal rainfall remains notoriously unpredictable, because it is determined by the interaction of numerous changing and competing factors, including: ocean currents and temperatures, sea surface temperature and wind conditions in the vast Indian and Western Pacific Ocean, phases of the El Nino Southern Oscillation in the equatorial Pacific, the Eurasian and Himalayan winter snow covers, solar energy output, and even wind direction and speed in the equatorial stratosphere some 30-50 km aloft.

Relying on computer climate models has one well-known side effect: Garbage in, gospel out. Current gospel certainly says CO2 rules the climate, but any role played by CO2 in monsoon activity is almost certainly dwarfed by other, major influences. Computer climate models have simply failed to confirm current climate observations, or project future climatic changes and impacts.

Both Indian and global monsoons have declined in strength and intensity over the last 50 years, and this reality largely contradicts climate model forecasts that say monsoonal rainfalls will increase. It is equally well known that climate models have been unable to replicate the decadal to multi-decadal variations of monsoonal rainfalls. Fred Kucharski and 21 other climate modellers challenge the alleged CO2-monsoon linkage. Using World Climate Research Programme climate model analyses, they conclude that “the increase of greenhouse gases concentrations has had little impact on the [observed] decadal Indian monsoonal rainfall variability in the twentieth century.” Perhaps the Indian scientists missed their report.

No climate models predicted the severe drought conditions for the 2009 Indian monsoon season - followed by the extended wetness of the 2010 season. The inability to foresee this 30-50% precipitation swing in most regions underscores how far we really are from being able to forecast monsoons, for next year, 2030 or the end of the century. Another recent analysis, by scientists from National Technical University in Athens, found that computer model projections did not agree with actual observations at 55 locations around the world. Computer forecasts for large spatial areas, like the contiguous US, were even more out of sync with actual observations than is the case with specific locations!

Ramesh says India hopes to offer a middle ground and present a less “petulant and obstructionist” perception during climate negotiations in Cancun. But if he believes the new report and claims of imminent forecasting ability will make this happen, we fear he is mistaken. “What-if” scenarios based on CO2-driven computer models are hardly a sound basis for negotiations, energy policies, agricultural planning or changed perceptions.

The impotence of current climate models is not surprising. As climate scientists, we know computer climate models are very useful for analysing how Earth’s complex climate system works. But models available today are simply not ready for prime time, when it comes to predicting future climate, monsoons or droughts. Our understanding of how weather and climate vary from year to year is still very immature, and it will be years (if not decades) before we resolve fundamental questions of how various forces interact to cause those changes.

Computer models still cannot accurately simulate or predict regional phenomena like the Indian summer monsoon rainfall. Even when model outputs agree with certain observations, we cannot be certain that the models did so for the right reasons. Considering the myriad factors that influence and alter weather and climate regimes, it is clear that climate models cannot make meaningful projections about future events, especially if they focus on the single factor of rising atmospheric CO2 levels.

Science and society will pay a very dear price, if political agendas continue to generate and legitimise false and pretentious computer outputs that have no basis in reality. How much better it would be if researchers focused on improving our ability to accurately forecast monsoons, droughts and other events just a few weeks or months in advance. That would really give farmers and others a chance to adapt, minimise damages and actually benefit from being better prepared.

Willie Soon is a solar physicist and climate scientist at Harvard-Smithsonian Centre for Astrophysics. Madhav Khandekar is a former research scientist from Environment Canada and served as an expert reviewer for IPCC’s 2007 reports.