Spectral binning of precomputed correlated-k coefficients^★

Laboratoire d’astrophysique de Bordeaux, Univ. Bordeaux, CNRS, B18N, allée Geoffroy Saint-Hilaire, 33615 Pessac, France
e-mail: jeremy.leconte@u-bordeaux.fr

Received: 27 July 2020
Accepted: 26 October 2020

Abstract

With the major increase in the volume of the spectroscopic line lists needed to perform accurate radiative transfer calculations, disseminating accurate radiative data has become almost as much a challenge as computing it. Considering that many planetary science applications are only looking for heating rates or mid-to-low resolution spectra, any approach enabling such computations in an accurate and flexible way at a fraction of the computing and storage costs is highly valuable. For many of these reasons, the correlated-k approach has become very popular. Its major weakness has been the lack of ways to adapt the spectral grid/resolution of precomputed k-coefficients, making it difficult to distribute a generic database suited for many different applications. Currently, most users still need to have access to a line-by-line transfer code with the relevant line lists or high-resolution cross sections to compute k-coefficient tables at the desired resolution. In this work, we demonstrate that precomputed k-coefficients can be binned to a lower spectral resolution without any additional assumptions, and show how this can be done in practice. We then show that this binning procedure does not introduce any significant loss in accuracy. Along the way, we quantify how such an approach compares very favorably with the sampled cross section approach. This opens up a new avenue to deliver accurate radiative transfer data by providing mid-resolution k-coefficient tables to users who can later tailor those tables to their needs on the fly. To help with this final step, we briefly present Exo_k, an open-access, open-source Python library designed to handle, tailor, and use many different formats of k-coefficient and cross-section tables in an easy and computationally efficient way.