Diagnostic Assessment and Potential Predictability of Regional and Large-Scale Wind Droughts in the Conterminous United States

As wind energy penetration increases, so too do the consequences of episodic wind droughts on wind power production, particularly when such droughts develop on large spatial scales and persist for a prolonged period. In this paper, a wind drought refers to anomalously low wind speed conditions at 100m that occur on the subseasonal to seasonal time scale and spatially on the synoptic to large-scale. The spatial domain considered is the conterminous U.S., with links between wind variability over land and the larger atmospheric circulation also considered. Using gridded wind field data from the Modern Era Retrospective-analysis for Research and Applications (MERRA), recurrent spatial patterns of anomalous 100m wind speed across the U.S. are first identified using empirical orthogonal function (EOF) analysis. The EOF analysis is applied on different timescales, ranging from daily to seasonal. As expected, the analysis reveals the important role of propagating synoptic systems on wind variability when performed using daily data. However, when performed on temporally aggregated data the results reveal recurrent, large-scale patterns in wind speed variability associated with wind droughts on the longer time scales of interest to this study. Canonical correlation analysis (CCA) is then utilized to identify associations between spatial patterns in 100m wind speed and the large-scale atmospheric circulation. For example, the wind drought in the western U.S. during the first half of 2015 is found to be associated with the episodic development of an anomalous ridging pattern over the western U.S. that in 2015 was also associated with severe drought conditions in California. In addition, statistical relationships between large-scale wind drought patterns and anomalous sea surface temperatures (SSTs) were examined via CCA. A statistically significant relationship between some large-scale wind drought patterns and contemporaneous SSTs suggests the former may be potential predictable.