When you ride a bicycle, the wheels spin and you can see it. The climate also has a rotation, but you can't see it with your eyes: it happens in the world of numbers, where each weather state is a point. Scientists have found a way to measure this rotation, as if they attached a speedometer to the climate. Probabilistic angular momentum is the name of the new tool that shows how fast and in which direction the climate 'spins' between different states.
El Niño is a warm current that comes every few years and changes weather around the world. It is tracked by two numbers: the water temperature in the ocean and the depth at which the water is warm. If you plot these two numbers on a plane, you get a point that moves over time. And this movement is not random: the point goes in circles, like the hand of a clock. The new tool measures exactly how strongly it spins.
The scientists took data on El Niño over 56 years and on another climate cycle that lasts a few weeks. For each day, they calculated a small rotation of the point on the plane. Then they summed all the rotations and saw that the distribution is asymmetric: most often there is no rotation, but sometimes it sharply appears in one direction. It's as if the speedometer usually showed zero, but occasionally the needle jerked to the right.
Previously, it was unclear how to find such rotation in real data. Now the tool exists, and it can be used to test climate models. If a model does not show rotation, it means it is missing something. Moreover, the same technique can be applied to other systems that constantly receive and give off energy, such as living organisms.
Probabilistic angular momentum has the same units as the diffusion coefficient: length squared divided by time. For El Niño, with its indices of temperature (°C) and depth (cm), the units become exotic: °C·cm/month.
The idea that climate oscillations are manifestations of hidden rotations in an abstract space echoes Isaac Asimov's novel 'The End of Eternity', where changes in reality are described as loops in a multidimensional state space.