Automatic Generation of Exposure Correction Factors for the Calculation of Potential Wind

The potential wind is the wind speed one would measure under ideal conditions, i.e. on a wind mast of 10 m height with only short grass and no obstacles or hills in the surroundings. It can be computed from the measured wind speed using direction-dependent exposure correction factors (ECFs), which are determined from the wind measurements themselves.

Climatology studies should be based on data that are representative of an area wider than the vicinity of the wind mast, so the data should exclude local effects (e.g. from nearby trees or buildings) and sensor changes or measurement site re-locations. If the “disturbance” of the measurement is not extreme, it can be compensated using ECFs (homogenisation).

Potential wind and ECFs help to answer a multitude of questions from different users, such as:
• Climatology: Is a change in wind speed over a 30-year period due to climate change or a change in local vegetation?
• Aviation: Should one expect significantly different wind speeds along an airport runway compared to what is measured?
• Inspection: Do station inspectors need to pay special attention to specific changes in the environment of a wind mast?
• Weather centre: Does a difference between forecasted and measured wind speed indicate a lower forecast quality?
• Construction sector: What extreme wind speeds should one assume when constructing buildings including primary water defence lines?

Essentially, there are two methods to compute ECFs: (1) the Gustiness method, based on the hourly maximum 3-second mean wind speed (gust) divided by the hourly mean wind speed and (2) the Sigma method based on 10-minute standard deviation divided by 10-minute mean wind speed. The latter method was proved to be superior in 2009 (Wever & Groen, 2009), but because 10-minute wind measurements were not stored before 2003, KNMI kept using the Gustiness method. Since this was a labour-intensive process, ECFs were only updated roughly every five years and therefore not always reflected most up-to-date changes. Now that KNMI has a substantial (and still growing) dataset, the Sigma method can be applied. Because this is already a major change in the method, we decided to make additional improvements that were not implemented before for continuity reasons, e.g. taking wind-direction dependence for reference surface roughness and height into account for coastal stations.

We developed, fine-tuned, tested and verified an algorithm that can automatically provide 10-minute ECF and potential wind speed updates based on the Sigma method. This allows us to give near-real-time insights to existing users, while also unlocking a broader range of users, including forecasters. At the same time, it reduces manual work load from months of intensive research every few years to occasional small support, advice and maintenance tasks. For technical reasons, only a monthly update will be produced operationally for now.

The data are to be published on the KNMI Data Platform (KDP) where they are readily available both to internal and external users. The old Gustiness-method dataset should be discontinued after a certain grace period but remain available as a static dataset (migrated to KDP). This transition requires clear communication with existing and new users, and this report shall contribute to that.