Analysing the uncertainties of reanalysis data used for wind resource assessment: A critical review

This review critically updates the state of the art on the uncertainties associated with reanalysis data directly used for wind resource assessment. Performances were analysed from 15 (9 global and 6 regional) reanalysis products applied on 322 very heterogeneous locations worldwide. Reanalysis scores were assessed by location type (offshore, coastal, inland and mountainous) and height above the ground (10-300 m). The results confirmed that reanalyses can predict wind resource pattern over time, while they are not similarly able over space. Reanalysis data (particularly by ERA5) are sufficiently reliable on offshore and flat onshore locations. Uncertainties are greater on mountainous and coastal sites, where wind speed and energy yield are significantly under- and over-estimated, respectively. On these locations, the use of higher-resolution regional products should be preferred over global datasets as they are better suited to resolving detailed terrain features as well as roughness and topographical discontinuities. Comparison between older and newer reanalysis versions revealed that the latter do not always outperform the former. In many cases, however, improvements by the newer products were found - rather than in accuracy itself - in pushing a given accuracy towards higher elevations, thus more suitably meeting the needs of modern wind power industry. In perspective, it is desirable that advances in forecasting models will enable current problems in assimilating surface land wind observations by global reanalyses to be overcome. Refinements in global datasets could also positively reflect on regional products.