Noise reduction methods in the analysis of near infrared lunar occultation light curves for high angular resolution measurements

Abstract

A lunar occultation (LO) technique in the near-infrared (NIR) provides angular resolution down to milliarcseconds for an occulted source, even with ground-based 1 m class telescopes. LO observations are limited to brighter objects because they require a high signal-to-noise ratio (S/N ∼40) for proper extraction of angular diameter values. Hence, methods to improve the S/N ratio by reducing noise using Fourier and wavelet transforms have been explored in this study. A sample of 54 NIR LO light curves observed with the IR camera at Mt Abu Observatory has been used. It is seen that both Fourier and wavelet methods have shown an improvement in S/N compared to the original data. However, the application of wavelet transforms causes a slight smoothing of the fringes and results in a higher value for angular diameter. Fourier transforms which reduce discrete noise frequencies do not distort the fringe. The Fourier transform method seems to be effective in improving the S/N, as well as improving the model fit, particularly in the fainter regime of our sample. These methods also provide a better model fit for brighter sources in some cases, though there may not be a significant improvement in S/N.