Mapping the surface of partially cloudy exoplanets is hard

Reflected light photometry of terrestrial exoplanets could reveal the presence oceans and continents, hence placing direct constraints on current long-term habitability these worlds. Inferring albedo map a planet from its observed curve is challenging because different maps may yield indistinguishable curves. This degeneracy aggravated by changing clouds. It has previously been suggested that disk-integrated spanning multiple days be combined to obtain cloud-free surface an exoplanet. We demonstrate this technique as part Bayesian retrieval simultaneously fitting for fixed time-variable overlying test approach synthetic data then apply it real observations Earth. find eight continuous are sufficient reconstruct faithful low resolution map, without needing make assumptions about cloud physics. For lightcurves with negligible photometric uncertainties, minimal top-of-atmosphere at location good estimate albedo. When applied Earth Polychromating Imaging Camera aboard DSCOVR spacecraft, our removes only small fraction attribute difficulty full-phase geometry short correlation length Earth-like climatology, hard do much better than cloud-averaged map. surmise removal will most successful imaged near quarter phase harbour large systems.