For nearly two centuries, the rain falling on Genoa has been doing more than filling streets, rivers and reservoirs. It has been quietly recording the changing behavior of Mediterranean precipitation extremes. Now, a new study based on the city’s exceptionally long observational record, spanning 1833 to 2022, argues that scientists may need far longer climate records than previously assumed to reliably detect whether extreme rainfall is becoming genuinely non-stationary. The central message is striking: for some precipitation indicators, records shorter than 150 years may not provide enough evidence to distinguish a persistent climatic shift from the powerful randomness of natural variability.
The finding matters because extreme rainfall is among the most difficult climate signals to interpret. Average precipitation can change gradually, but intense downpours are intermittent, localized and statistically rare. A year with several devastating storms may be followed by a long period of relative calm, even when the underlying climate is changing. Researchers therefore rely on probability distributions to estimate quantities such as the annual maximum daily rainfall or the amount of precipitation expected during a one-in-50-year event. If those distributions are assumed to remain constant over time, the analysis is described as stationary. If their average, variability or upper extremes evolve, the process is considered non-stationary.
In practical terms, detecting non-stationarity means proving that the statistical behavior of extreme rainfall has changed enough to rise above the background noise of weather. That is much harder than simply observing that recent years appear wetter or more damaging. A cluster of exceptional storms can occur by chance, particularly in a region influenced by complex interactions among the Mediterranean Sea, atmospheric circulation, mountainous terrain and coastal convection. The Genoa record offers an unusually powerful test because it covers approximately 190 years, far longer than the few decades commonly available for many weather stations. It allows researchers to examine how the length of an observational record affects the confidence of conclusions about changing precipitation extremes.
The study’s evidence points toward a fundamental limitation in conventional climate-extreme analysis. When researchers use shorter segments of a long record, statistical tests may fail to identify a real trend or may produce unstable conclusions about the direction and magnitude of change. The problem becomes especially acute for very rare events, because a short record contains only a small number of observations near the tail of the distribution. In statistical language, the tail is where uncertainty is greatest. A trend in the frequency or intensity of extreme rainfall can therefore remain hidden until the sample becomes long enough to include a sufficiently broad range of natural variability. According to the Genoa analysis, that threshold can extend beyond 150 years for some signals.
This does not mean that climate change cannot be detected using shorter records, nor that every region requires two centuries of observations before action is possible. Instead, the research highlights a distinction between detecting a change with high statistical confidence and making a sensible risk-management decision under uncertainty. Short records can reveal physical mechanisms, shifts in atmospheric circulation and changes in observed impacts. They can also be combined with climate-model simulations, reanalyses, historical documents and neighboring station records. But when the specific question is whether the statistical distribution of Mediterranean rainfall extremes has departed from a stable baseline, the study suggests that conventional tests may be underpowered unless the observational window is exceptionally long.
Genoa is particularly valuable for this purpose because of its geographical setting. The city lies at the meeting point of the Ligurian Sea and the steep relief of northwestern Italy, where moist marine air can be forced upward rapidly. Under favorable atmospheric conditions, this combination can generate intense rainfall over a short period. Such events are shaped by small-scale processes that are difficult for global climate models to represent directly. They are also sensitive to the exact position and speed of weather systems. The result is a record in which extreme values can fluctuate dramatically from one period to another, making it an ideal natural laboratory for testing the reliability of trend-detection methods.
The technical challenge extends beyond identifying a simple upward or downward trend. Non-stationarity can appear through changes in the location parameter of a probability distribution, which represents its typical level; through the scale parameter, which describes variability; or through the shape parameter, which controls the heaviness of the distribution’s tail. A climate record may show little change in ordinary rainfall while its most intense events become more variable or more probable. Researchers must also consider serial dependence, measurement practices, changes in instruments and the possibility that individual storms are influenced by different physical regimes. Each complication increases the amount of data needed to separate a persistent signal from statistical fluctuation.
The Genoa record therefore carries a message for engineers, planners and policymakers as much as for climate statisticians. Design standards for drainage systems, flood defenses, roads, bridges and urban development often depend on estimated return levels. If those estimates are calculated from a short, apparently stable record, they may underestimate the likelihood of future extremes or attach excessive certainty to a fragile number. Conversely, declaring every recent cluster of storms evidence of a permanent shift can lead to overconfident projections. A longer historical perspective does not eliminate uncertainty, but it reveals how large that uncertainty can be and helps decision-makers avoid treating limited observations as if they were complete descriptions of the climate.
The study also underscores the scientific value of maintaining long-running weather stations. In an era dominated by satellites, automated sensors and high-resolution models, a continuous ground-based record may appear old-fashioned. Yet the Genoa series demonstrates that observations collected across generations can answer questions that modern datasets alone cannot. Long records provide the statistical depth needed to examine rare events, test assumptions about stationarity and evaluate whether apparent changes persist across multiple climatic periods. They also create benchmarks against which newer instruments, remote-sensing products and model simulations can be assessed. Losing such records would mean losing irreplaceable evidence about how regional climate extremes behave over timescales relevant to infrastructure and society.
The broader implication is both cautionary and urgent. Mediterranean communities are already exposed to flash floods, landslides and rapidly escalating damage when intense rainfall strikes densely developed coastal areas. The Genoa findings do not offer a single universal number for how long every climate record must be, but they challenge the assumption that a few decades are automatically sufficient for detecting changes in rare precipitation extremes. In a warming world, scientists will need methods that combine long observations with physical understanding and model-based evidence, while clearly communicating the limits of statistical certainty. The oldest rainfall records may thus become some of the most important tools for navigating the future: not because they predict every storm, but because they reveal how easily short-term impressions can mislead us about a changing climate.
Subject of Research: Detection of non-stationarity and record-length requirements in Mediterranean precipitation extremes.
Article Title: Record-length requirements for detecting non-stationarity in Mediterranean precipitation extremes: evidence for periods greater than 150 years from the Genoa, Italy record (1833–2022)
Article References: Springer Nature article, DOI: 10.1007/s00382-026-08301-1
Image Credits: AI Generated
DOI: 10.1007/s00382-026-08301-1
Keywords: Mediterranean climate, precipitation extremes, extreme rainfall, non-stationarity, climate variability, Genoa, Italy, return levels, statistical detection, long-term weather records
