The Power of Time-Frequency Representations
Adrian Oncica, Cristiana Dumitrache, Vasile Mioc, Nedelia Antonia Popescu · AIP conference proceedings · 2007
Unlike most scientific disciplines, astronomical model‐building must typically rely on observations alone. No direct manipulation of the system or classical physical experiment is possible. Because of this, there has long been a close connection between astronomy and the study of time series. In fact the modern form of latter discipline arguably starts with Yule's analysis in 1927 of sunspot data series and the introduction of what we now call an ARMA model. In astrophysics we deal generally with inverse problems. Observations of some phenomena have to be understood even we have little knowledge on its driving mechanism or where exactly it operates. The problem is then to discover the nature of the underlying physics or geometry and how it produces the observed effects. For many signals in nature the frequency content changes in time as the physical processes producing the signal are time dependent. The tool for describing time‐varying spectrum is called time‐frequency analysis and saw significant progress in recent years. We intend to give a short review of the idea and its development and to point out qualities and drawbacks. Some examples from both long term and short term solar related data will be given. Finally we will focus on some long term Ulysses data.