Global Mapping of Surface Composition on an Exo-Earth Using Sparse Modeling

TL;DR

This paper introduces a sparse modeling approach for analyzing reflected light curves from exoplanets to map surface compositions, successfully distinguishing features like land, ocean, and clouds without prior surface knowledge.

Contribution

The study develops a novel sparse modeling method using $ ext{l}_1$-norm and Total Squared Variation regularization for surface mapping from light curves, enhancing resolution and accuracy.

Findings

Successfully inferred surface features from simulated data.

Identified cloud, ocean, and land components in real Earth data.

Improved geographic resolution over traditional methods.

Abstract

The time series of light reflected from exoplanets by future direct imaging can provide spatial information with respect to the planetary surface. We apply sparse modeling to the retrieval method that disentangles the spatial and spectral information from multi-band reflected light curves termed as spin-orbit unmixing. We use the $\ell_1$-norm and the Total Squared Variation norm as regularization terms for the surface distribution. Applying our technique to a toy model of cloudless Earth, we show that our method can infer sparse and continuous surface distributions and also unmixed spectra without prior knowledge of the planet surface. We also apply the technique to the real Earth data as observed by DSCOVR/EPIC. We determined the representative components that can be interpreted as cloud and ocean. Additionally, we found two components that resembled the distribution of land. One of the components captures the Sahara Desert, and the other roughly corresponds to vegetation although their spectra are still contaminated by clouds. Sparse modeling significantly improves the geographic retrieval, in particular, of cloud and leads to higher resolutions for other components when compared with spin-orbit unmixing using Tikhonov regularization.