Evidence that Some Reported Low-Frequency Solar Oscillations are Aliases

Sampling a continuous-time signal too slowly results in the phenomenon known as aliasing where high--frequency signals are folded into the low--frequency Nyquist band. Typically, one is interested in identifying oscillations (often from discrete modes) in the original signal by a frequency-domain analysis of the sampled signal. In addition to being at low frequencies, many of these oscillations have relatively low amplitudes and hence require long data spans to detect. In many instances the only available data is at daily or longer intervals. Consequently, making reliably physical inferences is difficult because one does not know if a given peak in a spectrum is real or an alias of a higher frequency component. In a previous paper (Moghtaderi, Thomson and Takahara (2010)), we proposed an algorithm to detect aliasing, and approximately unfold them. The algorithm uses data from two or more locations where the observations are offset in time and chooses the frequency where the phase of the coherence most closely agrees with that predicted from the time delay between the two stations. Many discrete, low-frequency oscillations have been identified in solar and terrestrial data. We describe a frequency-domain analysis of eighteen years of daily RSTN and Penticton solar noon flux data, collected from January 1989 to December 2006 that confirms the presence of many of these oscillations. However, by using data from five observatories, we are able to detect aliasing and estimate the original frequencies. We show that it is likely that many of the low-frequency oscillations reported in the literature are aliases of higher frequency signals. To support this assertion, we also analyzed geomagnetic data sampled at one hour and one minute intervals and observe that the frequencies of lines in these data sets agree with the unaliased (corrected) low frequency modes from the solar radio data.