You ought to be familiar with the concept of the leap day—the addition of a 29th day in February ever four years, as just occurred in 2024—but you may not have given as much thought to the humbler (but troublesome) leap second.
Tying our timing to celestial clockwork served us well in the past, but the advent of more precise methods based on atomic standards revealed some awkward mismatches. It turns out that the rotation of the Earth, and therefore the length of a day, varies just enough that the behavior of the heavens eventually drifts slightly out of sync with the atomic clock. This isn’t significant enough to give you an excuse for arriving late to an appointment, but it does present a thorny problem for precisely defining and tracking time.
Coordinated Universal Time (UTC) is based on the atomic standard, while UT1 time is based on the rotation of the Earth. Since 1972, there have been 27 manual additions of a leap second to UTC time to prevent it from ever drifting more than 1 second from UT1 time. These take the form of an extra second added to the end of the last day of June or December. That means that all systems that rely on timing have to be ready to smoothly handle these unusual events.
A new study published in Nature by Scripps’ Duncan Agnew tries to disentangle the major factors affecting Earth’s rotation—and finds that we may have a technical problem our on our hands in the next few years.
Leap seconds don’t occur on a predicable schedule because the processes that make them necessary don’t, either. Variations in Earth’s rotation are a function of the conservation of angular momentum and movements of mass. Fluids on Earth’s surface can move around somewhat freely, slightly changing the distribution of mass around the globe. Just as a figure skater can spin faster or slower by raising their arms above their head or reaching them outward, movements of fluids toward the poles or the equator will affect our planet’s spin rate.
The same is true for changes in the distribution of solid rock, although such a thing may be harder to picture. But large additions or subtractions of mass from an area (such as growing glacial ice or melting it away) cause the surface of the Earth to be depressed downward or rebound upward. This “isostatic adjustment” is a significant factor for the Earth’s rotation. A large area of the high-latitude Northern Hemisphere is still adjusting to the loss of the massive ice sheets of the last ice age, for example.
Separately, the Earth’s liquid outer core plays a large role. The inner core and outer solid portion (mantle and crust) of the planet can exchange some momentum―with one speeding up slightly at the expense of the other. This complex process is not understood well enough to be predictable. It swings back-and-forth over years and decades.
And as a final long-term factor, it’s also true that Earth’s rotation has been very gradually slowing down over its history due to tidal friction—the Moon’s continuous gravitational tug on the Earth.
Watching the clock
We precisely track the Earth’s rotation, and the length of a day, over time. The greater the difference between the actual length of a day and exactly 86,400 seconds, the faster UTC time is drifting from UT1 time. In this study, Agnew analyzed this record and compared it to our best estimates of the factors affecting Earth’s rotation.
After subtracting the effects of glacial isostatic adjustment and tidal friction, modern climate change has to be accounted for. The melting of glacial ice—converting solid mass mostly nearly the poles into liquid that spreads out in the oceans—has accelerated as the global temperatures increase.
That will continue accelerating in the coming decades, to an extent depending on future greenhouse gas emissions. But based on the change so far, and an extrapolation of this through 2045, the effect on Earth’s rotation can be estimated.
With this also subtracted, the complex interaction between the core and the outer solid Earth should now dominate the record. That has been trending towards shortening days since the mid-1970s, with an opposing effect from sea level rise helping to obscure it recently.
While the future of this trend can’t be predicted, extrapolating each of these effects shows that we could have an interesting problem to solve soon. The core-outer Earth trend is stronger than the effect of sea level rise, and the overall effect on Earth’s rotation has returned to where it was in 1972. Unless this pattern reverses, we’ll be entering a phase where negative leap seconds are needed for the first time. And by extrapolating the current trends, Agnew found that we’ll need the first intervention within the next five years.
Leap seconds cause considerable headaches, and this has only gotten worse over time as we rely on precise computerized systems. The manual insertion of an extra second can crash software, corrupt data, and complicate scientific analyses. Recent leap seconds have taken down large websites, caused flight delays, and caused hiccups in financial systems. As a result, tech companies have called for a less disruptive alternative to the leap second, particularly with the possibility of subtracting a leap second on the horizon—an operation we have yet to experience.
There is momentum behind these calls for change. The General Conference on Weights and Measures voted in 2022 to loosen the synchronization between UTC and UT1 by 2035, allowing them to drift apart by more than the current one-second standard and making leap seconds less frequent. But they’d have to decide on what that new standard would be. As this new study points out, a decision may be needed much sooner than 2035 in order to avoid our first run-in with a negative leap second.