Posters

Abstract

Analyzing wind behavior at a specific location is crucial when planning the construction of a wind farm. Not only the extension, but also the quality and availability of the data at the location significantly impacts the accuracy of the analysis. In cold regions, high altitudes, or areas with extreme temperature fluctuations, data gaps are an inevitable challenge due to equipment failures or environmental factors. Filling these gaps often requires utilizing data from nearby measuring masts or external reanalysis sources. However, in certain scenarios, those sources may not be available. When this occurs, the only viable solution is to fill the missing data using other levels from the same mast. Filling these gaps is critical, and the process often involves the use of Measure-Correlate-Predict (MCP) methods. While there are many MCP approaches available, this study focuses on comparing two of the most widespread methods and the combination between them: correlation by hour, by sector, and by a combination of sector and hour. The reliability of MCP methods is not universal; and the proposed study seeks to analyze the effects of site characteristics, such as terrain complexity and wind directional patterns, on their performance. The study was conducted using data from more than 500 measuring met masts situated in diverse regions and terrain types, each with extensive campaigns of recorded data. Data gaps were randomly introduced at one mast level, and the fabricated gaps were filled using the three aforementioned MCP methods. The interpolated results were then compared with the original measured data to evaluate accuracy. An essential aspect of the study was determining the minimum dataset required to establish a reliable correlation when the correlation by sector and hour was to be used. Our analysis revealed the minimum discrepancies were found when using a specific threshold of data points per sector and hour for valid correlations. Fewer points forced filling with a very low representativeness, which led to inaccuracies, while higher thresholds restricted the applicability, forcing reliance on single-parameter correlations (sector-only or hour-only), which also increased discrepancies. Interestingly, the combined sector-and-hour correlation method proved highly effective, delivering accurate results, and was only very rarely outperformed by simpler methods under specific site conditions. These rare cases occurred when the site’s convective effects completely overpowered the orographic influences, or vice versa, leading to conditions where simpler methods outperformed the combined sector-and-hour approach. The determination of the most effective method for each mast depended significantly on its location. Results showed that the optimal method differed depending on regional and terrain characteristics. Interestingly, masts within the same geographical area exhibited similar behaviors, suggesting a consistent pattern influenced by location. In conclusion, the study demonstrates that a mast's behavior in data correlation and interpolation is strongly influenced by its location and terrain complexity. By understanding the regional and terrain-based factors influencing MCP performance, wind data analysis can be optimized for improved wind farm planning and development, improving both the accuracy and confidence of wind resource assessments for wind farm development.