Principal components of the surface wind field in the United States: A comparison of analyses based upon wind velocity, direction, and speed

Abstract Three statistical summaries of the 1975 surface wind field in the United States are computed using principal components analysis (PCA). Separate analyses of wind velocity and direction are made with vector‐based PCA, and the scalar (zonal and meridional) components of wind are described using a scalar‐based PCA. Comparisons indicate that results of the three PCAs are not equivalent. Vector PCA yields the most physically comprehensive components because the velocity observations are not decomposed into direction and speed, or into zonal and meridional coordinates a priori. Directional and scalar‐based PCAs appear similar but since they only summarize one coordinate of velocity at a time, they represent incomplete descriptions of the wind field. Vector‐based PCA then is the most appropriate of the three for investigating wind fields or other vector‐valued data sets.