Correlation Correlation Measurement in Measure-Correlate-Predict Techniques

摘要

In this study, data from fifteen pairs of reference and target sites are used to compare predictions of mean wind speed and wind distribution with measured values. The predictions are obtained using linear, matrix and parabolic-based correlations within an MCP context. These are also compared to predictions from the modified Weibull Scaling method, in which the Weibull parameters at the target and reference site during the concurrent period and at the reference site during the historic period are used to derive the Weibull parameters at the target site during the historic period. Results are then compared based on both the Pearson correlation, which assumes a linear correlation between the data, and the Spearman correlation, which is a more general rank correlation, between the site pair. The methods which are most successful at predicting the mean wind speed are not the most successful at predicting the wind distribution and as a result, energy yield, demonstrating that a comparison of mean wind speed is insufficient to establish the accuracy of an energy yield prediction.