Extreme value analysis for estimating 50 year return wind speeds from reanalysis data

摘要

This paper presents new methods for enabling the effective quantification of risk for the wind power industry. It focuses primarily on the challenges stemming from the limited availability of wind speed time series of sufficiently long duration for observing extremes. As a solution, the report investigates the use of reanalysis data from atmospheric dynamical models as a proxy for the short records of actual wind speed time series that are typically available. By developing an innovative approach for calibrating the reanalysis data, it is shown that superior estimates of the 50 year return level may be achieved using the reanalysis data. On the basis of 45 years of actual and reanalysis data collected at Schiphol airport, it is shown that this new approach is superior to using the actual data in all situations where less than 20 years of actual data are available. The improvements in the 50 year return level estimates for one, two and three years of actual data are 32.41%, 31.29% and 33.64%, respectively.