ISSN:
1572-9052
Keywords:
Time series
;
Bayesian approach
;
outlier detection
;
smoothing
;
nonlinear modeling
Source:
Springer Online Journal Archives 1860-2000
Topics:
Mathematics
Notes:
Abstract In the measurements of VLF electric fields with the Pioneer Venus spacecraft in sunlight, spin synchronized signals often dominate over the naturally generated emissions. We present a method to separate natural emissions from the several possible sources of noise. Our major objective by this method is not to remove all spin modulation, but to effectively subtract the background noise caused by the identifiable noise sources. Examination of the data shows that the background spin synchronized noise is quite sensitive to ϑ(n), the angle between the sense axis and the solar direction. We model the observed data asy(n)=w(n)t(n)f(ϑ(n))+x(n), wheref(ϑ) represents the phase response of the background noise andx(n) is the estimated natural emissions.t(n) andw(n) are the long-term trend component and time- and phase-independent component of the intensity of the background noise, respectively. The method to decomposey(n) is based on the Bayesian approach which has been recently applied to various inversion problems such as nonstationary time series modeling and image reconstruction. In this procedure, the estimated parametersw(n),t(n),f(ϑ), andx(n) can be determined automatically. We will describe the Bayesian scheme and its application to the Pioneer Venus VLF electric field data.
Type of Medium:
Electronic Resource
URL:
http://dx.doi.org/10.1007/BF00773508
Permalink
