Revisiting Spitzer transit observations with Independent Component Analysis: new results for the GJ436 system

We analyzed four Spitzer/IRAC observations at 3.6 and 4.5 {\mu}m of the primary transit of the exoplanet GJ436b, by using blind source separation techniques. These observations are important to investigate the atmospheric composition of the planet GJ436b. Previous analyses claimed strong inter-epoch variations of the transit parameters due to stellar variability, casting doubts on the possibility to extract conclusively an atmospheric signal; those analyses also reported discrepant results, hence the necessity of this reanalysis. The method we used has been proposed in Morello et al. (2014) to analyze 3.6 {\mu}m transit light-curves of the hot Jupiter HD189733b; it performes an Independent Component Analysis (ICA) on a set of pixel-light-curves, i.e. time series read by individual pixels, from the same photometric observation. Our method only assumes the independence of instrumental and astrophysical signals, and therefore guarantees a higher degree of objectivity compared to parametric detrending techniques published in the literature. The datasets we analyzed in this paper represent a more challenging test compared to the previous ones. Contrary to previous results reported in the literature, our results (1) do not support any detectable inter-epoch variations of orbital and stellar parameters, (2) are photometrically stable at the level 10e-4 in the IR, and (3) the transit depth measurements at the two wavelengths are consistent within 1{\sigma}. We also (4) detect a possible transit duration variation (TDV) of 80 s (2 {\sigma} significance level), that has not been pointed out in the literature, and (5) confirm no transit timing variations (TTVs) >30 s.