On the role of gaps in measurement data and filling strategies with respect to AEP calculations for offshore wind farms

Christine Martens1 ,3, Bilke Engelbrecht2 ,4, Martin Doerenkaemper 1, Bernhard Stoevesandt1, Julia Gottschall2
1Fraunhofer IWES, Oldenburg, Germany, 2Fraunhofer IWES, Bremerhaven, Germany, 3Fachhochschule Kiel, Kiel, Germany, 4WKN AG, Husum, Germany


The offshore wind climate is generally characterized by strong winds and other rough conditions such as wave height and sea spray. This imposes strong demands not only to offshore wind turbines but also to wind measurement installations like buoys and met masts. The majority of these installations is accessible by boat only, a very few allow helicopter support. Thus, the failure of measurement devices or their electricity supply can result in gaps of several days in the measurement time series, if the platform is not accessible due to weather condition.

Within this study the impact of these gaps in measurement data is investigated with a focus of their role on annual energy production (AEP). The seasonal impact of these gaps is analyzed as well. The study showed that there is generally a strong seasonal impact but this impact may vary depending on the wind conditions of a year.

In a second step these gaps are filled with uncorrected as well as corrected mesoscale simulation data. The results indicated that the filling of the gaps reduces the error in the mean wind speed and AEP by more than 80%. Finally a real case study will be presented with a failure of a measurement device at a met mast. An extension of the measurement period is compared to the filling methodologies presented in this study.

The study also states the cost of the required simulations and the correction algorithm.



After an analysis of the impact of random gaps on AEP and mean wind speed the gaps in the measurement data are filled by means of mesoscale model simulations (Skamarock et al., 2008). These simulations were performed with a set-up that has recently widely been used for offshore wind studies (e.g. Dörenkämper et al., 2015).
To correct for systematic errors in the simulation data and provide an uncertainty estimation, the Analog Ensemble Methodolgy is applied (Delle Monache et al., 2013). This method searches for analogous simulation data, in historical overlapping time frames, where measurement and simulation data are available. On the basis of the best fitting analogues a statistical mean and deviation is calculated to replace the missing value in the time series.


Gaps in measurement data generally show a strong dependency on the season. By filling these gaps in the data by means of uncorrected mesoscale simulations, the mean absolute percentage error (MAPE) of the mean wind speed was reduced to a tenth of the value without the filled gaps. An application of the Analog Ensemble Methodology afterwards leads to another reduction of about 40% with respect to the annual mean wind speed.


The novel method presented in this study presents an opportunity of filling gaps in measurement data by mesoscale simulations corrected with information from historical data. In contrast to other methods the Analog Ensemble Method (AEM) does not only provide corrected mean wind speeds but also additional information about the uncertainty of the filled value. The AEM allows for filling any given gappy time series can in a smooth way.


The impact of gaps in offshore measurement data on AEP calculations will be discussed. In addition the study presents a novel method for filling these gaps.