Calendar date ±14 days in each year; hourly interval endpoints converted to local time. Each complete morning has one mean temperature, summed rain and mean scalar wind. Variable-specific missing mornings are excluded, never filled.
Typical conditions in the area from a reanalysis model (Copernicus ERA5-Land for temperature, about 9 km cells; ERA5 for rain and wind, about 25 km), not a measurement in the street. City-centre mornings usually run 1–3 °C warmer than the model cell, temperatures are moved to the start’s height where the cell sits more than 100 m higher or lower, a rainy morning has at least 1 mm (the model drizzles more often than a rain gauge), and wind along streets differs from a 25 km cell. Not a climate normal, past race-day record, forecast or individual pace adjustment.
There are 232 eligible mornings across 8 years. Percentiles use linear interpolation. We require at least six years, 80% complete mornings and ten mornings in every year for each variable. These are coverage gates, not proof that the sample predicts race-day weather.
Nearby days in a year and hours in a morning are correlated. Rain counts use complete mornings and a 1 mm threshold over the entire selected window. Different variables can have different counts. No missing hour is treated as dry or filled from another station.
The published 2027-04-04 start is 09:00 local. The chosen window is a clock-time scenario, separate from your goal pace or training schedule. Organizer start reference.
Reanalysis cell near the start (48.0°, 7.8°) is at 316 m, 3.2 km from the reference start. The grid cells holding the course start: temperature from ERA5-Land (48.0°, 7.8°), rain and wind from ERA5 (48.00°, 7.75°). A model of the area, not a measurement in the street. The cell is within 100 m of the start's height (252 m), so no height correction is applied.
These recent model summaries are not the existing TerraClimate 1991–2020 monthly normals: monthly day/night averages cannot supply a race-hour distribution or rain-window count. No climate-trend comparison is made.
Coverage by year and hourly values
- 2018: temperature 29, rain 29, wind 29 complete mornings.
- 2019: temperature 29, rain 29, wind 29 complete mornings.
- 2020: temperature 29, rain 29, wind 29 complete mornings.
- 2021: temperature 29, rain 29, wind 29 complete mornings.
- 2022: temperature 29, rain 29, wind 29 complete mornings.
- 2023: temperature 29, rain 29, wind 29 complete mornings.
- 2024: temperature 29, rain 29, wind 29 complete mornings.
- 2025: temperature 29, rain 29, wind 29 complete mornings.
- 09:00–10:00: 3.7–14.9 °C; median 8.9 °C; 232 observations.
- 10:00–11:00: 4.9–16.1 °C; median 9.9 °C; 232 observations.
- 11:00–12:00: 6.1–17.5 °C; median 11.1 °C; 232 observations.
- 12:00–13:00: 6.9–18.8 °C; median 12.2 °C; 232 observations.
- 13:00–14:00: 7.5–19.7 °C; median 12.6 °C; 232 observations.
- 14:00–15:00: 7.7–20.1 °C; median 13.1 °C; 232 observations.
Hourly timestamps are UTC interval endpoints, converted to Europe/Berlin with daylight saving time. We use 2018 onward, the period of the station summaries. Source interval documentation.
Annual source review; changes to the race date, start time, course revision or observations require regeneration and review. Method: reanalysis-mornings/1. Derived summaries by TrailSplits, with Copernicus attribution and licence.