Wind resource assessments produce hundreds of thousands of measurements every year. Before determination of the wind power density, a function of the velocity cubed, those values are screened to remove erroneous data points. Typical categories of erroneous data are sensor malfunction, tower shading, and icing. Identification of tower shading is a well established procedure dependent on the mounting direction of the sensor booms. Most instrument malfunctions are clearly extended flat lines and typically only affect one sensor at a time. Sensor icing of anemometers and directional vanes, on the other hand, can be subtle and affect more than one sensor simultaneously and can require an experienced evaluator’s assessment. Designation of icing results in the removal of lower velocity data. If too few points are removed the wind velocity will be underestimated, while if too many points are removed the wind velocity can be exaggerated, both of which can have a significant influence on the power density due to the cube effect. And experts frequently disagree. Much of this disagreement is driven by the difference between rule based approaches and operator judgment approaches. A comparison of different screening approaches for icing is described in this work. Three different rule-based approaches are compared against a visually-based expert determination that combines multiple sensors, including temperature, humidity, directional vanes and anemometers with several rule based approaches. Relative impacts of different approaches can affect from 1.09% for a visually-based expert approach to 5.03% for a rule-based standard deviation approach.

This content is only available via PDF.
You do not currently have access to this content.