Researchers analyze historical modeling outputs to assess seasonal climate predictions
Bathtub rings show the impact of sustained drought on Lake Mead, Nev. In a new study, researchers compare historical modeling outputs to gain a better understanding of different types of seasonal forecasting skills. Credit: 4kodiak/iStock/Getty Images Plus/Getty Images
Source: Journal of Geophysical Research: Atmospheres

Seasonal climate predictions offer important information about anticipated regional conditions for periods extending beyond standard, long-range weather forecasts. Products summarizing these data, like the National Oceanic and Atmospheric Administration’s 3-month outlooks, provide valuable guidance for governments and natural resource managers.

Seasonal predictions may be either deterministic—a “point” forecast that states an event will occur at a specific place and time—or probabilistic, an indication of the likelihood that an event will occur. The accuracy of probabilistic forecasts depends upon two characteristics: the reliability, which is how consistently the predicted occurrence probabilities of an event match the corresponding observed occurrence frequencies, and the resolution, the degree to which the observed frequencies differ from the long-term climatological frequency. Recently, researchers discovered that a relationship exists between the probabilistic resolution and the deterministic forecasting skills but have been unable to fully explain it.

Now Yang et al. have conducted a detailed analysis of the relation between these skills to further investigate this puzzle. The team first confirmed that under specific theoretical conditions, a monotonic relationship exists between deterministic and resolution forecasting skills. The researchers then analyzed historical general circulation model (GCM) outputs generated by the European ENSEMBLES project for the period spanning 1960 to 2005. The findings indicate that the established relationship can also be confirmed using GCM forecast data.

The results, which offer a fresh perspective on two of the most intensely studied measures of forecasting skills, may have implications for improving our understanding of the properties of seasonally averaged variability in the atmosphere. The research also highlights the importance of not only evaluating the accuracy of seasonal predictions from a probabilistic perspective but also improving our understanding of how each skill is related to other types of forecasts. (Journal of Geophysical Research: Atmospheres,, 2018)

—Terri Cook, Freelance Writer


Cook, T. (2018), Evaluating the accuracy of seasonal climate predictions, Eos, 99, Published on 12 July 2018.

Text © 2018. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.