Abstract
Principal component analysis is applied to analyze the horizontal component of geomagnetic data for the Panzhihua Ms 6.1 earthquake. We investigate temporal variations in eigenvalues and find that only the first principal component has good correlation with the Ap index for which the cross-correlation correlation R is larger than 0.6, which may imply solar-terrestrial activity. Both the second and third principal components show clear daily variation, being high during work hours and low at night and on weekends. The mean eigenvalue of the third component at night (00:00-04:00 LT) increased about 40 days before the earthquake and returned to normal 10 days before the earthquake. These features are likely to be correlated with the earthquake.