[Previous] | [Session 53] | [Next]
P. A. Sturrock (Stanford U.), D. O. Caldwell (UC Santa Barbara), J. D. Scargle (NASA Ames Research Center), G. Walther (Stanford U.), M. S. Wheatland (U. Sydney)
The Super-Kamiokande Collaboration has released two datasets suitable for time-series analysis: one packaged in 10-day bins and, more recently, one packaged in 5-day bins. We have analyzed both datasets by a likelihood power-spectrum procedure that is to be preferred over the more usual Lomb-Scargle procedure since it takes account of all of the available data, whereas the Lomb-Scargle procedure uses only a fraction of the data. Analysis of these two datasets yields power spectra that are similar but have some notable differences. Comparative analysis of these two datasets shows that the significant differences are due to aliasing. Since each dataset represents highly regular sampling, a modulation at frequency vM will be accompanied by alias periodicities at frequencies \vert vT - vM \vert and at vT + vM, where vT is the sampling frequency. As indications of real oscillations, we have looked for peaks in the two power spectra which are strong in both, but for which the power in the 5-day power spectrum is larger than that in the 10-day power spectrum. This plot identifies three peaks of special interest. One of these may be attributed to modulation associated with solar rotation, and the other two may be attributed to modulation associated with an internal r-mode oscillation. These periodicities appear to be statistically significant. We thank the Super-Kamiokande consortium for making these datasets available. This work was supported by NSF grant AST-0097128 and DOE grant DE-FG03-91ER40618.
[Previous] | [Session 53] | [Next]
Bulletin of the American Astronomical Society, 36 #2
© YEAR. The American Astronomical Soceity.