Vineyard Frost Risk: Assess Any Site's Frost Window

In April 2021, growers across France lit tens of thousands of candles between their rows overnight and still lost an estimated 2 billion euros of production to a single run of frost nights. The vines were fine in January at -10 °C. They were devastated in April at -4 °C. That asymmetry is the whole subject of frost risk: it is not about how cold a site gets, it is about when.

This guide covers how frost actually damages vines and fruit trees, the difference between air and ground frost, how terrain creates frost pockets, and how to read any site's frost window from ten years of climate data before planting a single vine.

Why Spring Frost Is the One That Hurts

A dormant grapevine tolerates roughly -15 to -20 °C. But once buds burst, the young shoots are mostly water and die at about -2 °C. Warmer springs push bud break earlier while the date of the last frost moves more slowly, so in many regions the danger window has widened, not shrunk. The same logic applies to orchards: apple blossom at full bloom is killed at about -2 °C, and a single night erases the crop that year.

So the numbers that matter for a site are: when does the last spring frost typically strike, when does frost return in autumn, and does that window overlap bud break and flowering for what you want to grow.

Air Frost, Ground Frost, and Why +2 °C Can Still Burn Shoots

Weather stations measure temperature at 2 meters. Air frost means that reading hits 0 °C. But on clear, calm nights the ground radiates heat to the sky and runs colder than the air above it, so ground frost (freezing at the surface, where buds on low-trained vines live) occurs with 2 meter readings as warm as about +2 °C. A site whose April minimum reads +1.5 °C is not safe, it is a ground frost site.

The two mechanisms of frost also behave differently. Radiation frost (calm, clear nights, cold air pooling) is the common one and the one terrain and protection can fight. Advection frost (a cold air mass moving in with wind) overwhelms candles and wind machines alike; if a site sees regular advection frost inside the growing season, the answer is variety choice or a different site, not equipment.

Reading a Site's Frost Window: Three Real Examples

The numbers below come straight from PixelGust's frost hazard, which derives each site's window from the coldest reading each calendar month recorded over 2015-2025 (ERA5 reanalysis):

The table tells the story growers know from experience. Chablis is famous for spring frost losses precisely because air frost persists into April and ground frost into May, exactly when its early-budding Chardonnay is most vulnerable; hence the candles. The Douro's schist slopes see only marginal ground frost by April, one reason its steep terraces carry vines so reliably. Michigan's fruit belt compresses everything into a 4-month frost-free season, and its May air frosts are why growers there plant late-blooming apple varieties and site orchards on ridges.

Mendoza, Argentina makes the same point in the southern hemisphere: frost possible until October (its spring), returning in April, a High rating that matches the frost fans dotted across its vineyards.

Terrain: Where Frost Actually Settles

On radiation frost nights, cold air behaves like water: it flows downslope and pools in low ground. Two parcels a few hundred meters apart can differ by 3-5 °C at dawn. The rules of thumb:

• Mid-slope beats valley floor. The classic vineyard position lets cold air drain past and through the rows. Valley floors and hollows are frost pockets.
• Check what blocks the drainage. Woods, walls, and dense hedges below a parcel dam the cold air flow and create a pocket above them.
• Aspect trades frost for other risks. East-facing slopes thaw fast after frost nights (which can worsen damage to frozen tissue); slope and aspect data shows what each parcel faces.
• Elevation cuts both ways. Higher sites are cooler on average but often sit above the pooled cold air on radiation nights, a phenomenon called thermal belts.

Gridded climate data at ~10 km resolution captures a site's regional frost regime, not these local effects. The honest workflow is climatology first (does the frost window overlap the growing season at all), then terrain analysis for drainage, then a winter of data loggers on the shortlisted parcels.

What to Do With the Answer

• Choosing between sites: compare frost windows and frost-free season length. A month less frost exposure is worth more than most soil differences.
• Choosing what to plant: match bud-break timing to the window. Late-budding varieties buy 2-3 weeks of safety at frost-prone sites.
• Budgeting protection: wind machines run 30,000+ euros per unit; candles cost thousands per hectare per night. Whether that capital is needed at all falls out of the frost window analysis, and pairs with growing degree days for the full climate suitability picture.
• Due diligence on land: a frost assessment belongs in any environmental profile of agricultural land before purchase.

How to Check Frost Risk on PixelGust

1. Open the dashboard at pixelgust.com/app and click your site, or enter its coordinates.
2. Open the Hazards panel. The Frost Risk row shows the rating (None to High, based on whether frost intrudes into the growing season) and the frost window in plain language.
3. Dig into the months in the Weather panel's Historical view: monthly minimums for 2015-2025 show exactly how cold each month has gone.
4. Check the terrain in the Terrain panel: elevation, slope, and aspect for the cold-air drainage picture.
5. Save it as a report: the Property Climate Report bundles frost risk, the monthly climate history, terrain, and the other hazards into one dated PDF.

Data Sources and Method

Frost ratings are derived from ERA5 reanalysis (ECMWF) monthly minimum 2 meter temperatures, 2015-2025, at ~10 km resolution. Air frost is a monthly minimum at or below 0 °C; likely ground frost is a minimum at or below +2 °C. Terrain data is Copernicus DEM at 30 m. Frost pockets and other microclimate effects below the grid scale are not resolved and always warrant on-site verification for planting decisions.