Posters

Abstract

Site wind measurements carried out for proposed wind farms typically cover a duration of up to a few years, requiring that they are put into a longer-term context. ECMWF ERA5 is a global Reanalysis dataset which typically provides good correlation with site measurements, making it a useful long-term data source both onshore and offshore. A further important criterion for selecting an appropriate long-term reference source is consistency. Including inconsistent data in the definition of the long-term period can introduce significant bias into the resulting long-term wind resource estimate. Typical methods used to define consistency rely on comparison between multiple independent data sources. However, many wind projects are proposed in regions where good quality meteorological networks are not available. As a result, proving long-term consistency of reanalysis data can be challenging, leading to increased uncertainty. Here, a clear, confident and globally applicable approach to defining reference data consistency enables a significant reduction in that uncertainty. The ERA5 data assimilation method ingests a wide range of observations, including radiosondes, buoys, stations, measurements and satellite data. These observations have changed over time for a variety of reasons: improvements in technology, changes in the coverage of observational networks, observation frequency and changes to satellite missions to name a few. Due to these considerations, the risk of inconsistencies increases as we go further back in time and the rate of this change differs between regions. In this presentation, the ensemble spread parameter is used as a proxy for uncertainty to investigate the consistency of ERA5 reanalysis wind speed data across different regions. Historical consistency inferred globally from this parameter will be presented along with observation of how this has changed from the early satellite era through to the current state of the art observation systems. This contrasts the relatively high consistency in northern Europe with the lower consistency observed in important emerging wind markets with fewer historical measurements such as Asia and south America. The author will present a method that fully unlocks the potential of parameters provided as part of the ERA5 dataset to identify the most appropriate long-term reference period for a given site location, balancing length of dataset with consistency; a significant advance from the current commonly used method of selecting a generic start date for all locations. The audience will gain an understanding of the impact that using inconsistent data can have on energy yield assessment results (in terms of both potential bias and uncertainty), which ERA5 parameters can be used to infer consistency of ERA5 data, and how historical changes are dependent on region. It will be clear how the proposed approach minimises the risk of including inconsistent data in an analysis, so providing the best possible long-term wind speed estimates for energy yield assessment.