Wavelet estimation for hidden periodic components in spatial series.

抄録

An overlapped wavelet method is proposed to detect the number and locations of the hidden periodicities in two dimensionally indexed random fields, by checking if the empirical wavelet coefficients of periodogram have significantly large absolute values across fine scale levels. The magnitudes of the amplitudes are also estimated using wavelet coefficients of smaller scale levels than that for the detection of periodicities. The strong consistency of the estimators is established. Some numerical examples are given to test the performance of our method.