Principal component analysis (PCA) was applied to 182 half-hour runs containing time series of turbulent wind velocity and temperature measured in the convective atmospheric surface layer. A field experiment with four sonic anemometers on the vertices and one in the centroid of a square (with sides of 80 m) was performed to obtain the necessary dataset. Physical explanations of the most important eigenvectors are presented. Two of the major principal components (PCs) identify the variance in wind speed along and across the background wind direction. Always, one major PC accounts for the presence of large-scale thermal activity: periods with higher (lower) temperatures coincide with lower (higher) wind speeds, convergence (divergence) in the wind fields, and upward (downward) movements. As an application, variance in the velocity fields was expressed in terms of horizontal divergence and vertical vorticity. These can be derived directly from the eigenvectors when PCA is combined with a planimetric method. Using the PC that identifies thermal activity, it is found that the magnitude of divergence increases and the magnitude of vorticity decreases when atmospheric conditions become more unstable. It is found that the (absolute) ratio between vorticity and divergence scales with a function of the friction velocity divided by the convective vertical scaling velocity. Both kinematic parameters are larger for updrafts than for downdrafts. It is concluded that PCA can be a useful tool to distinguish variance of thermal and nonthermal origin and in the estimation of the kinematics of dominant flow fields.