Soil breathes. The top layer of the earth takes in moisture after rain, then releases it into the air, much like lungs inhale and exhale. This inhalation and exhalation happens everywhere on the planet, but its rhythm is not uniform: in some places it is fast and steady, in others irregular and deep. The SMOS satellite listened to this breathing for eight years, measuring soil moisture every three days from an altitude of 750 kilometers. The recorded signal was not just noise: it contained clear rhythms that older methods could not separate.

The new method, named ROCK PCA, works like an experienced physician listening to a chest. Standard principal component analysis (PCA) could only hear one loud tone: the annual cycle, which explains 70.6% of all variability. But within that loud breathing, quieter rhythms were hidden. To hear them, the method first shifts the phase of the signal (Hilbert transform), as if flipping the recording to the other side, then compresses the data using the Gram matrix to speed up computations. Instead of comparing millions of points with each other, it compares only time series, which takes thousands of times less effort.

Next, the method adds nonlinearity via a kernel, a function that measures similarity between points not along a straight line but along a curve. Finally, it rotates the coordinate axes (promax rotation) so that each rhythm becomes distinct and interpretable. As a result, four main rhythms emerged from the overall soil breathing. The first is the annual cycle, like a steady inhale and exhale every 12 months. The second and fourth are harmonics, that is, overtones of this cycle with periods of 6 and 4 months: they are related to how tropical rains shift north and south. The third rhythm turned out to be the most surprising: a slow oscillation with a period of about 4.5 years, which does not match the seasons.

This slow rhythm intrigued scientists the most. They compared it with the El Niño index (ENSO4), a measure of how warm the water becomes in the central Pacific Ocean. The correlation between them reached 0.8, which is very high for natural data. This means that when El Niño warms the ocean, soil moisture rises in some regions and falls in others. The map shows: in Eastern Europe, southeastern South America, and the southern United States, the ground becomes wetter, while on the eastern coast of Australia, in northern South America, and near the Gulf of Guinea, it becomes drier.

Before this method, such connections were hard to see because standard analysis mixed all rhythms into a single blur, and computations took too long. Now we can monitor soil breathing in real time and understand which rhythm currently dominates. It is like the difference between hearing the noise of a crowd and distinguishing individual voices. The new method can be applied not only to soil moisture but also to ocean temperature, vegetation state, and other global datasets. Soil breathes, and now we can hear every breath it takes.