University of Minnesota
Characterizing Signal Detectability in Gappy KELT Light Curves Using Wavelet Analysis
Abstract
dc:description.abstractWe present an investigation into the feasibility of detecting stellar rotation signals in gappyground-based photometric data using wavelet-based methods, with the goal of assessing the suitability of such data for convolutional neural network (CNN)–based rotation period extraction. Using light curves from the KELT North telescope, we inject synthetic rotation signals and evaluate their recoverability with both Morlet-type [3, 4] and Weighted Wavelet Z-transform (WWZ) wavelets [8, 12]. We find that idealized sinusoidal signals injected into KELT light curves become detectable with the WWZ wavelet for peak-to-peak amplitudes down to approximately 0.5% of the flux baseline. However, rotation signals in some stellar populations, particularly late-type M dwarfs, can exhibit amplitudes as small as 0.08% [11]. An analysis of the KELT flux distribution shows that the noise is well described by a Gaussian profile. By constructing gap-free synthetic light curves through random sampling of this distribution at approximately the KELT cadence, we successfully recover rotation signals with peak-to-peak amplitudes as small as 0.08% with the Morlet2 wavelet. These results demonstrate that, while Gaussian noise alone does not prevent the recovery of low-amplitude rotation signals, the severe time gaps inherent to ground-based observations are the dominant limiting factor. Consequently, while ground-based datasets can be used to recover stellar rotation periods for a subset of stars, gap-free observations from space-based missions, or coordinated ground and space-based campaigns, remain the preferred approach for robust rotation studies and machine-learning applications.
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Laine, Matt
Subjects
dc:subject × 12- Irregularly sampled light curves
- Kilodegree Extremely Little Telescope (KELT)
- Signal detectability
- Stellar rotation
- Time-series
- Wavelet analysis
- Master of Science in Physics
- Department of Physics and Astronomy
- Swenson College of Science and Engineering
- University of Minnesota Duluth
- Plan As (thesis-based master's degrees)
- Master of Science
Rights
- Language dc:language.iso
- en
Identifiers
dc:identifier.*- Dc Identifier Citation
- Laine, Matt. (2026). Characterizing Signal Detectability in Gappy KELT Light Curves Using Wavelet Analysis. Retrieved from the University Digital Conservancy, https://hdl.handle.net/11299/280352.