Weather prediction hits a theoretical 129-day ceiling

For decades, meteorologists have grappled with the question of how far weather predictions can reliably extend. A recent study published in Advances in Atmospheric Sciences offers a compelling answer: even under perfect conditions, there exists a fundamental limit to weather forecasting, which is approximately 129 days. This limit emerges from a new approach that reframes the question by focusing on the atmospheric energetics rather than prediction errors.

Previously, researchers had attempted to determine this limit by studying how errors grow over time. However, lead author Dr. Wei Zhang, a climate scientist at the University of Miami and the NOAA Cooperative Institute for Marine and Atmospheric Studies, points out that this method has limitations. Without understanding how errors smaller than those in today's forecasts might behave, it's challenging to claim that very-long-range forecasts can be skillful.

To address this gap, the research team, including co-author Dr. Zoltan Toth, decided to explore the problem from a different perspective. They asked whether there might be a fundamentally new way to approach the perennial question of predictability. By refining the key question under ideal conditions—with precise knowledge of the initial state of the atmosphere, governing dynamics, and future macroscale boundary conditions—they sought to establish the ultimate limit of weather prediction.

Instead of focusing on prediction errors, the team turned to the basics: the energetics of the atmosphere. They reasoned that if the initial state were known exactly, the dynamics of the atmosphere would preserve that knowledge, leading to perfect forecasts indefinitely. The only exception, they assumed, would be the quantum-scale uncertainty introduced into the atmosphere through the phase of photons in the constant influx of solar radiation. This uncertainty, which is unknown, ultimately determines the limit of weather predictability.

By examining the atmospheric energy cycle, the team concluded that solar radiation, which powers all atmospheric motion, would eventually reach every molecule. As the solar energy propagates through the atmosphere, the uncertainty associated with the unknown phase of incoming photons would blur all memory of the initial state, rendering precise prediction impossible.

This "energy turnover point"—when the atmospheric energy has been sufficiently mixed by solar radiation—marks the boundary beyond which weather prediction becomes unreliable. Considering the total energy of the atmosphere, the incoming flux of solar radiation, and observational uncertainties, the authors determined that the likely limit of internal weather predictability is 129 ± 7 days.

Today's forecast skill, which averages about 14 days, remains far below this theoretical ceiling. However, the remaining time beyond this limit can be divided roughly equally between periods of genuine skill extension and periods of marginal skill. For instance, under ideal conditions, a 5-day forecast might theoretically be extended to about 62 days, after which guidance would become low-confidence.

The study's findings not only establish a theoretical ceiling for weather prediction but also provide forecasters with a clear target for future advancements in this field.