Wind farms · France · nightly migration forecast

In the Netherlands the offshore wind farms are stopped on the nights of heavy migration, with a forecast built on the weather radars. This page points the per-city forecast of phase 3, which works at any point without a radar, at the wind turbines, and adds the other half of the decision: what stopping would cost that night.

Zones. The turbines mapped in OpenStreetMap are grouped into cells of 0.5° (about 56 × 38 km); a cell with at least 10 turbines is a wind zone, with its forecast point at the centroid of its turbines. The alert level is cut, as for the cities, on the percentiles of the zone's own record around the same date: high is above its 75th percentile and very high above its 90th.

Cost of stopping. The forecast wind at 100 m over the night hours, through a generic power curve (nothing below 3 m/s, rated power from 12 m/s, cut-out at 25 m/s), gives the share of rated power the zone would produce. Below 25% stopping is cheap; from 50% it is expensive. The two halves cross in a fixed table: very high and cheap → stop; very high and moderate → stop at the peak hours; very high and expensive, or high → watch; anything below → run. It is a way of ranking nights, not an operating order.

Why so strict. Over the 2021-2026 seasons on the Peninsula (ERA5 reanalysis, 151 zones), this table would have stopped about 7.5 nights per season and zone for 3.1 % of the night output, covering 8.8 % of the migration the model predicts; stopping on every very high night would have cost 12.8 %. It comes out this cheap because the heavy nights tend to be calm ones. The nearest radar, where there is one within 75 km, measured a busy night on 67 % of the suggested stops against 24 % by chance: hindcast and radar check.

Zone-nights where stopping is suggested

Full forecast