Statistical analysis of record-breaking temperatures

Author

In brief

Climate change is usually communicated through averages, but the extremes are what leave the deepest marks on societies and ecosystems. Record-breaking events offer a simpler, more communicative and statistically robust way to evaluate deviations from a stationary climate, because the probability of a record in an independent identically distributed series is known exactly and does not depend on the underlying distribution. This article reviews the hypothesis tests that follow, and applies them to forty Spanish temperature series.

“The accelerated pace of record-breaking events underscores the urgency of implementing both mitigation and adaptation strategies.”

What the article covers

• Under stationarity the probability of a record at time t is exactly 1/t, whatever the distribution.
• Combining four types of record, forward and backward, upper and lower, to build more powerful tests.
• Daily maxima from 40 Spanish stations, 1960 to 2021, with Benjamini–Yekutieli adjustment for multiplicity.
• The effect of warming is neither spatially uniform nor homogeneous across the year: summer is the most affected season.

How to cite

Ana C. Cebrián. “Statistical analysis of record-breaking temperatures.” vISIon: The ISI Magazine Vol. 1 No. 1, October 2025, pp. 11–14.