Two pendulums hang from the same spring. One swings left, while the other is already returning from the right. They always lag each other by half a period. If you watch the first pendulum, you can tell in advance where the second will be half a period later. The Indian monsoon and global El Niño are linked in much the same way.

The monsoon brings rain to India every year. El Niño, every few years, warms a large patch of the Pacific Ocean and disrupts weather worldwide. Previously, it was thought that the monsoon could be predicted only six months ahead: beyond that, chaos, that is, random eddies and fluctuations, ruins everything. But 140 years of weather records showed that if you shift El Niño data forward by 18 months, they match monsoon data. The match is very strong: a coefficient of 0.84, where 1 would mean a perfect match.

The scientists built a model from two coupled equations. Each system in the model resembles a pendulum with its own internal dynamics. The coupling is arranged so that when one system strengthens, the other weakens after a delay, and then vice versa. This is similar to how an increase in the number of hares leads, after some time, to an increase in the number of foxes. The model showed the same as the observations: without a shift, the correlation is almost zero, but with an 18-month shift, it becomes very high. This means that patterns in one system repeat those in the other after 18 months.

Now it is clear why the monsoon can be predicted a year and a half ahead, even though chaos limited forecasts to six months. The secret is not in overcoming chaos, but in synchronization: one system shows in advance what will happen to the other. This approach could be used to find other pairs of climate systems linked with a delay. Then forecasts could become longer and more reliable without complex calculations.