Principles
- Every number has a source, listed at the foot of each page.
- A dash means not captured yet, never "no rule".
- One climate method everywhere, so regions can be compared fairly.
- Comparisons are chosen by the data, never by hand, and described on neutral scales (Jones, Winkler, Huglin).
How the climate figures are made
- Each region is represented by a few vineyard sample points at real vineyard heights.
- For each point, hourly weather since 1990 comes from Copernicus ERA5-Land (about 9 km grid), turned into daily values and adjusted from the grid's average height to the vineyard's height (−0.65 °C per 100 m).
- Reanalysis smooths out narrow valleys and rain shadows, so each profile is anchored to a nearby official weather station: month by month, temperatures are shifted and rainfall scaled so the 1991–2020 averages match the station (moved to the vineyard's height). Year-to-year variation still comes from ERA5-Land.
- From the daily values come the indices: growing-season temperature, Winkler degree days, Huglin, cool night index, day–night range, hot days, frosts and rainfall in the season, at veraison and at harvest. Seasons run April–October in the north and October–April in the south, labelled by harvest year.
Anchor stations
ERA5-Land validation (first full run, 26 October 2026)
ERA5-Land normals (1991–2020, after the lapse-rate correction to vineyard height) compared with station normals we already hold. Winkler GDD (April–October, base 10 °C).
| Station (normal 1991–2020) | Station GDD | ERA5-Land, nearest sample point | Difference |
|---|---|---|---|
| Dijon (Côte d’Or) | ~1324 | Côte d’Or profile 1289 | −3% |
| Bolzano (254 m) | ~1901 | Bolzano-Gries ~1407 | −26% |
| Merano-Gratsch | ~1620 | Merano ~1047 | −35% |
| Brixen/Bressanone (585 m) | ~1418 | Novacella ~860, Chiusa ~828 | −40% |
| Siena (320 m) | ~1991 | Chianti profile 1684 (vineyards 280–500 m) | −15% (partly real: height, urban warmth) |
Findings
- Gentle terrain is fine. Côte d’Or is within a few per cent; Bordeaux, Baden, Sonoma and Willamette look plausible (not yet checked against stations).
- Alpine valleys fail. A 0.1° cell (~9 km) in Alto Adige averages valley floor and mountains (cell mean heights 550–1580 m against vineyards at 240–700 m). Reanalysis cannot resolve the warm valley floors, and the lapse correction (+4 to +6 °C at some points) is too uncertain to fix it. It also scrambles the order between points: Caldaro (430 m) comes out warmer than Bolzano-Gries (260 m).
- Central Otago fails. Hottest-month maximum ~19–20 °C against ~24–25 °C observed around Alexandra; annual rain ~1,000–1,300 mm against roughly 350–450 mm in the basin. The rain shadow is not resolved. GDD 495 for the profile is unusable.
- Marlborough is too cool and too wet. Profile GDD 944; Wairau hottest-month maximum 21.8 °C against ~24 °C at Blenheim; rain ~1,000 mm against ~650 mm.
- The lapse shift uses a 90 m DEM mean of the cell, not ERA5-Land’s own model orography, which adds error in steep terrain.
Approach
- Done: absolute level from station normals for every profile (station anchoring).
- Done: vintages keep ERA5-Land’s year-to-year variation, bias-corrected month by month against the station normal (delta method).
- To do: use ERA5-Land’s own orography (geopotential invariant) for the lapse correction.
- Partly done: each profile has a confidence level (station-anchored, or graded by the size of the height correction). Terrain roughness of each cell is still to add.
- Done: station normals added for all eleven profiles (see caveats below for the provisional ones).
Results after station anchoring (normals 1991–2020)
| Profile | Anchor station | GDD, ERA5-Land only | GDD, anchored | GST °C | Winkler | Growing-season rain, mm (before → after) |
|---|---|---|---|---|---|---|
| Baden | Freiburg | 1231 | 1277 | 15.6 | Ib | 521 → 521 |
| Bordeaux | Bordeaux-Mérignac | 1651 | 1763 | 18.2 | III | 455 → 487 |
| Burgundy (Côte d’Or) | Dijon-Longvic | 1289 | 1321 | 15.9 | Ib | 544 → 456 |
| Alto Adige · over 500 m | Brixen-Vahrn/Bressanone-Varna | 982 | 1386 | 16.2 | Ib | 795 → 573 |
| Alto Adige · 220–300 m | Bozen/Bolzano | 1684 | 1912 | 18.9 | III | 738 → 536 |
| Alto Adige · 300–500 m | Meran-Gratsch/Merano-Quarazze | 1511 | 1555 | 17.1 | II | 807 → 582 |
| Chianti Classico | Siena | 1684 | 1894 | 18.8 | III | 495 → 445 |
| Central Otago | Alexandra | 496 | 696 | 12.7 | Ia | 717 → 241 |
| Marlborough | Blenheim | 944 | 1166 | 15.5 | Ib | 524 → 335 |
| Sonoma (Russian River and Sonoma Coast) | Santa Rosa, CA | 1463 | 1574 | 17.3 | II | 140 → 137 |
| Willamette Valley | Salem McNary Field, OR | 1138 | 1206 | 15.4 | Ib | 390 → 310 |
Caveats
- Siena normals are interim (a climate aggregator, whole degrees). Official monthly normals are on ISPRA’s per-station site (valori-climaticinormali.isprambiente.it), which could not be reached from here.
- Freiburg (DWD) has mean temperature only; daily range and rainfall stay uncorrected ERA5-Land. Fix: compute from DWD daily data for station 1443.
- Alexandra has only 10 years of data in 1991–2020, and a standard lapse rate probably over-cools Central Otago’s slopes at night (strong valley inversions). Central Otago’s anchored GDD (~700) is lower than commonly quoted figures; treat it as provisional.
- Blenheim and Alexandra come via Figure.NZ/Wikipedia republishing NIWA; confirm against NIWA’s XLSX summaries and record station numbers.
- One anchor station per profile: within-profile differences between points come only from elevation.
Vintage check: anchored ERA5-Land against station-derived vintages
Vintage indices calculated independently from the Bolzano and Brixen station records (monthly data, in the Alto Adige record) compared with the anchored ERA5-Land profiles, 1991–2025:
| Growing-season temp. r | GDD r | Growing-season rain r | |
|---|---|---|---|
| Bolzano vs low band | 0.94 | 0.94 | 0.71 |
| Brixen vs high band | 0.85 | 0.86 | 0.68 |
Means agree within 0.3 °C. Year-to-year temperature variation is well captured; rainfall less so (local showers), so per-vintage rain should be presented with a little caution.