Case study · Agricultural forecasting

Frost alerts a grower can act on

Global weather models run on a 25 km grid — they cannot see the radiative cooling that freezes a specific orchard. The system calibrates ensemble models (ECMWF IFS & AIFS, DWD ICON-EU) against the on-site station sensor, blends them by likelihood, and turns the result into frost forecast up to 5 nights ahead, enough time to plan protection, not just react. The model can predict all frost nights in the dataset;

What the service includes

On-site sensor installation — temperature and humidity in your orchard, the local truth every forecast is calibrated against.
Calibrated forecast system — global ensemble models, bias- and spread-corrected for your site and blended into one verified probability.
Real-time monitoring — a live dashboard, refreshed every forecast cycle.
Alert system — WhatsApp/SMS the moment frost probability crosses your cost/loss threshold, hours before the minimum.

Case study site: Argamasilla de Alba (SIAR CR07) — grapevine, critical temperature −1.5 °C.

Verified against reality

The blue band is the model’s own 90% confidence interval, checked against two seasons of on-site data: nine nights out of ten the observed temperature falls inside it.

Observed temperatures vs the calibrated forecast with its 90% uncertainty band for the week of the 2026-03-30 frost — the observations stay inside the band and dip below the critical temperature on the frost night — One cold week from the archive: observed vs the forecast issued the morning before, with the frost night caught below −1.5 °C.

Explore the live dashboard — every metric, updated each forecast cycle →