Key points are not available for this paper at this time.
Principal component (PC) analysis performed on irregularly spaced data can produce distorted loading patterns. We provide an example to demonstrate some distorted patterns which can result from the direct application of PC analysis (or eigenvector analysis, factor analysis, or asymptotic singular decomposition) on irregularly spaced data. The PCs overestimate loadings in areas of dense data. The problem can be avoided by interpolating the irregularly spaced data to a grid which closely approximates equal-area.
Karl et al. (Sun,) studied this question.