The study combines hourly and daily observations from Météo-France and the French electricity company EDF across a combined network of 163 stations. The records span the northern and southern French Alpine provinces, where complex topography intersects competing Atlantic and Mediterranean climatic influences. For the climatology, the researchers used recent observations from 1980 through 2022 or 2023. The trend analysis relied on EDF records extending as far back as 1952, including 28 stations with more than 45 years of data and a longest series of 71 years. The long-term network was concentrated in mountain areas, with few EDF stations in the Rhône valley and along the Mediterranean coast, an important limitation when comparing trend patterns with the broader stationary climatology.

Quality control was particularly important because the older EDF material contains only weekly maximum precipitation totals for fixed durations of 1, 2, 3, 6, 12 and 24 hours. For recent hourly series, the researchers compared sums of 24 hourly readings with quality-controlled daily totals and scaled each day’s hourly values when necessary to reproduce the recorded daily precipitation. Days with missing daily observations or discrepancies exceeding 25 percent were invalidated, and visual inspection of seasonal maxima led to the removal of one station with suspicious winter values. Instrumentation and data-processing methods changed during the historical record, including a transition from mechanical to digital registration, so the authors assumed that individual time series were broadly homogeneous while treating known irregularities at the earliest stations as a limitation.

For each station, the team applied extreme value analysis to monthly, seasonal and annual precipitation maxima using a generalized extreme value distribution. A 20-year return level represents rainfall with a 1-in-20 annual chance of being exceeded under the fitted model, making it a measure of a rare, potentially high-impact event rather than the most intense value directly observed. The non-stationary model keeps distribution parameters unchanged before 1985 and allows them to evolve linearly afterward. Separate models permitted trends in the location parameter, the scale parameter or both, with the best-fitting form selected by the Akaike information criterion. A bootstrap procedure resampled each maximum series 500 times and assessed whether the 5 to 95 percent confidence interval for the estimated change included zero.

Because estimating trends from records of only a few decades can be unstable, the model borrowed observations from nearby stations. For each target station, annual maxima from surrounding sites were normalized by their respective means, combined and rescaled to the target station. The neighborhood radius was 0.3 degrees, adding an average of 3.9 neighboring stations, although sites near the regional margins had fewer neighbors. Tests using radii of 0.2 and 0.4 degrees produced broadly similar results. The technique makes regional patterns more robust but assumes that nearby stations share a statistically homogeneous precipitation distribution, an assumption that may fail across steep terrain or distinct weather regimes. The approach also does not explicitly account for topography.

The baseline climatology reveals sharply different seasonal signatures for daily and hourly rainfall. Daily 20-year return levels were generally below 70 millimeters per day from February through August and rose above 200 millimeters per day in parts of Alpes-Maritimes from September through January. The Rhône valley and Mediterranean coast also stood out during the wetter autumn months. Hourly return levels followed a different annual cycle: they were lowest in February, generally below 15 millimeters per hour, and highest in late summer. Estimates reached about 75 millimeters per hour in western Alpes-de-Haute-Provence in July. The warm-season peak is consistent with localized thunderstorms and greater atmospheric moisture and instability, while the dry Mediterranean summer produces exceptionally low totals in the south. Hourly extremes were also higher along southern and western slopes and in the pre-Alps than in the drier inner Alpine zone.

Large-scale circulation helps explain the contrast between daily and hourly events. Atlantic flows dominated northern daily rainfall maxima throughout the year. In the south, Mediterranean and Atlantic influences alternated, with Mediterranean flows especially important from September through November and during the wider seasonal analysis. Daily annual maxima closely resembled autumn maxima, consistent with the importance of Mediterranean storms. Hourly events showed a less robust relationship to the daily weather-pattern classification, reflecting their more localized formation, although anticyclonic conditions became more influential in the south from May through August. The study used an existing classification of eight large-scale weather patterns and grouped three Atlantic and three Mediterranean flow types, together with northeast circulation and anticyclonic conditions.

Daily trends were generally between minus 10 and plus 10 millimeters per day per decade, but their seasonal and regional structure was coherent. The northern Alps showed significant increases in May and decreases in June and September. In the south, daily extremes decreased in August and September but increased from October through December. Seasonal trends were negative in winter and summer and positive in spring and autumn. Annual return levels in the south rose by as much as 24 millimeters per day per decade, equivalent to an estimated increase of up to 50 percent from 1990 to 2020, whereas northern annual trends reached minus 8 millimeters per day per decade, or about minus 20 percent over the same comparison. The strongest southern increases coincided with stronger Mediterranean circulation, while the opposite northern pattern was consistent with changes in Atlantic influence, although the observational study alone cannot establish causation.

Hourly trends were less consistent and statistically significant at fewer stations. Northern estimates were usually centered near zero, apart from relatively small negative September trends. In the south, trends leaned positive from April through June and from October through December, with slight negatives in August and September. November was the clearest exception: a spatially coherent positive pattern reached 3.5 millimeters per hour per decade in the south, corresponding to an estimated increase of up to 100 percent between 1990 and 2020. Annual hourly trends were also centered near zero in the north and only slightly positive in the south, with significant estimates at just two southern stations having opposite signs. A possible northeast-southwest gradient appeared in summer and annual results, with stronger values in the high Alps than in lower southwestern terrain, but the researchers did not present it as a definitive regional pattern.

The physical causes of the hourly pattern remain hypotheses rather than demonstrated mechanisms. The authors suggest that warming may increase the energy available for convection in high mountain areas, potentially aided by shorter snow persistence and associated albedo changes, while increasing aridity may reduce moisture for local recycling in lower areas. Seasonal shifts may also matter: the widespread decline in September daily extremes may indicate a later onset of autumn rainfall, while increases in November are consistent with that timing change. The study does not directly test these explanations because the EDF rain-gauge stations lack temperature measurements, and its hourly records cannot resolve subhourly behavior. More fundamentally, return-level estimates depend on record length, sampling and measurement quality, while the model assumes linear change after 1985 and statistical homogeneity within neighborhoods. Overall, the results show that daily and hourly hazards cannot be treated as interchangeable in the French Alps: circulation changes strongly organize daily extremes, whereas short-duration events reflect a more complicated balance among moisture, energy, convection and topography.

Subject of Research: Observed climatology and trends of daily and hourly precipitation extremes in the French Alps

Article Title: Climatology and trends of observed daily and hourly extreme precipitation in the French Alps

Article References: Berghald, S., Blanchet, J., Blanc, A., & Penot, D. (2026). Climatology and trends of observed daily and hourly extreme precipitation in the French Alps. Hydrology and Earth System Sciences, 30(20), 6421-6442. https://doi.org/10.5194/hess-30-6421-2026

Image Credits: AI Generated

DOI: 10.5194/hess-30-6421-2026

Keywords: French Alps, extreme precipitation, rainfall extremes, hourly rainfall, daily rainfall, extreme value analysis, Mediterranean storms, climate trends, flash floods, weather patterns, Climatology, trends