[Paper Review] Molecfit: A general tool for telluric absorption correction II. Quantitative evaluation on ESO-VLT X-Shooter spectra

Wolfgang Kausch, S. Noll. A. Smette+7 more|arXiv (Cornell University)|Jan 28, 2015

Astronomy and Astrophysical Research13 references180 citations

TL;DR

This paper evaluates Molecfit, a software tool that corrects telluric absorption in ground-based astronomical spectra using synthetic transmission spectra derived from atmospheric models, eliminating the need for time-consuming telluric standard star (TSS) observations. It demonstrates that Molecfit achieves superior correction quality compared to the classical TSS method by avoiding continuum distortion, intrinsic spectral features, and noise from TSS data, with high accuracy across diverse observing conditions and signal-to-noise ratios on ESO-VLT/X-Shooter data.

ABSTRACT

Context: Absorption by molecules in the Earth's atmosphere strongly affects ground-based astronomical observations. The resulting absorption line strength and shape depend on the highly variable physical state of the atmosphere, i.e. pressure, temperature, and mixing ratio of the different molecules involved. Usually, supplementary observations of so-called telluric standard stars (TSS) are needed to correct for this effect, which is expensive in terms of telescope time. We have developed the software package molecfit to provide synthetic transmission spectra based on parameters obtained by fitting narrow ranges of the observed spectra of scientific objects. These spectra are calculated by means of the radiative transfer code LBLRTM and an atmospheric model. In this way, the telluric absorption correction for suitable objects can be performed without any additional calibration observations of TSS. Aims: We evaluate the quality of the telluric absorption correction using molecfit with a set of archival ESO-VLT X-Shooter visible and near-infrared spectra. Methods: Thanks to the wavelength coverage from the U to the K band, X-Shooter is well suited to investigate the quality of the telluric absorption correction with respect to the observing conditions, the instrumental set-up, input parameters of the code, the signal-to-noise of the input spectrum, and the atmospheric profiles. These investigations are based on two figures of merit, I_off and I_res, that describe the systematic offsets and the remaining small-scale residuals of the corrections. We also compare the quality of the telluric absorption correction achieved with moelcfit to the classical method based on a telluric standard star. (Abridged)

Motivation & Objective

• To assess the performance of Molecfit in correcting telluric absorption in high-resolution ground-based spectroscopy without requiring additional telluric standard star (TSS) observations.
• To quantify the impact of observing conditions, signal-to-noise ratio, instrument setup, and input parameters on the accuracy of Molecfit's correction.
• To compare Molecfit’s correction quality directly with the classical TSS-based method, identifying key limitations of the latter such as continuum distortion and noise amplification.
• To validate the robustness of Molecfit under varying atmospheric conditions, including low water vapor and low S/N regimes.
• To demonstrate the generalizability of Molecfit across different instruments and observing sites through flexible parameterization and atmospheric profile adaptation.

Proposed method

• Molecfit uses the LBLRTM radiative transfer code and the HITRAN line database to compute synthetic telluric transmission spectra based on atmospheric parameters derived from fitting narrow wavelength ranges in the science target spectrum.
• The atmospheric profile is initialized using GDAS meteorological data, and the fitting process adjusts for wavelength grid distortions via adjustable polynomials (e.g., Chebyshev) to improve alignment with observed features.
• Two figures of merit, $I_{\mathrm{off}}$ (systematic offset) and $I_{\mathrm{res}}$ (residual noise), are used to quantitatively evaluate correction quality across the full U to K band spectrum.
• The method fits only narrow, feature-free regions of the spectrum to avoid contamination from strong molecular bands or intrinsic object features, ensuring stable atmospheric parameter retrieval.
• Input parameters such as the degree of freedom in wavelength correction and the initial atmospheric profile are tuned to minimize fitting errors and maximize robustness.
• The approach is validated using archival ESO-VLT/X-Shooter spectra covering the full visible and near-infrared range, with comparisons to both TSS-based corrections and independent atmospheric measurements from microwave radiometers.

Experimental results

Research questions

• RQ1 How accurately does Molecfit correct telluric absorption across the full U to K band in X-Shooter spectra under varying observing conditions?
• RQ2 To what extent does Molecfit outperform the classical telluric standard star (TSS) method in terms of continuum fidelity and residual noise?
• RQ3 How sensitive is Molecfit’s correction quality to signal-to-noise ratio (S/N) and the choice of fitting regions in the spectrum?
• RQ4 How robust is Molecfit to errors in the initial atmospheric profile, particularly under dry or extreme conditions?
• RQ5 Can Molecfit reliably correct low S/N spectra, or is a TSS observation still required in such cases?

Key findings

• Molecfit achieves a highly effective removal of telluric absorption features across the full U to K band, with $I_{\mathrm{off}}$ and $I_{\mathrm{res}}$ values indicating minimal systematic offsets and low residual noise.
• The method significantly outperforms the classical TSS-based correction by avoiding continuum distortion, intrinsic spectral features, and noise amplification from the TSS observation.
• Fitting only a few narrow, feature-free regions across the spectrum is sufficient to achieve high-quality correction over the entire wavelength range, especially when these regions are spread to improve wavelength calibration and line spread function correction.
• The fitting algorithm is robust to variations in the initial atmospheric profile, including temperature, pressure, and water vapor content, though PWV estimates are slightly overestimated in very dry conditions—without degrading correction quality.
• Molecfit performs reliably even on low S/N data, though correction quality degrades for extremely low S/N; in such cases, using a TSS for fitting is a viable alternative.
• The method is adaptable to various instruments and observing sites, with only minor parameter adjustments needed, and can be applied using non-ESO-compliant FITS headers by adding custom metadata in the parameter file.