The computational principle of the Morlet wavelet power spectrum W(t, k) of a meteorological time-series p={P(t), t=1, n}is intuitively illuminated using the geometric analysis in phase space, and a scheme of Monte Carlo significance test for W(t, k) is presented in this paper, that is convenient and simple because the generation of the random series is unrelated to the theoretical distribution of the analyzed series. Significance test is performed for wavelet power spectrum of the intensity index series of 100 hPa South Asia High in June, the results show that the scheme can sufficiently reveal the temporal evolution of interannual and interdecadal oscillations of p, thus it is of practical application value.