The Advantages of Global Photometric Models in Fitting Transit Variations

Estimation of planetary orbital and physical parameters from light-curve data relies heavily on the accurate interpretation of Transit Timing Variation (TTV) measurements. In this letter, we review the process of TTV measurement and compare two fitting paradigms-one that relies on making transit-by-transit timing estimates and then fitting a TTV model to the observed timings, and one that relies on fitting a global flux model to the entire light-curve data set simultaneously. The latter method is achieved either by solving for the underlying planetary motion (often referred to as "photodynamics"), or by using an approximate or empirical shape of the TTV signal. We show that, across a large range of the transit S/N regime, the probability distribution function of the mid-transit time significantly deviates from a Gaussian, even if the flux errors do distribute normally. Treating the timing uncertainties as if they are distributed normally leads, in such a case, to a wrong interpretation of the TTV measurements. We illustrate these points using numerical experiments and conclude that a fitting process that relies on a global flux fitting, rather than the derived TTVs, should be preferred.