Nacelle anemometer

Every turbine carries an anemometer and a wind vane on the nacelle roof. The controller uses them to decide when to start, how to yaw, and how to pitch. They are also the wind measurement that arrives in SCADA data, which makes them the default input to any analysis built on the turbine's own record.

The problem is their position. They sit downwind of the rotor, in flow the rotor has already slowed and disturbed, and behind a nacelle that generates its own wake. The reading is therefore systematically biased and the bias varies with operating state, which is why manufacturers apply a nacelle transfer function to convert measured speed into an estimate of free-stream wind. That function is itself a calibration, and it drifts.

The consequence is circularity that has to be handled explicitly. If a turbine's expected output is computed from a wind speed the turbine itself reported, a sensor fault can hide a performance problem: the machine underproduces and simultaneously reports less wind, so the pair looks consistent. This is one of the standard ways a real loss stays invisible for years.

Breaking the circle means bringing in a reference the machine does not control - a met mast, reanalysis data, or the neighboring turbines' measurements of the same weather. Cross-checking a machine's own anemometer against its neighbors is one of the more reliable ways to find a drifting sensor, and it is a routine part of loss attribution.