The SNR of a transit
David Kipping
Abstract
ABSTRACT Accurate quantification of the signal-to-noise ratio (SNR) of a given observational phenomenon is central to associated calculations of sensitivity, yield, completeness, and occurrence rate. Within the field of exoplanets, the SNR of a transit has been widely assumed to be the formula that one would obtain by assuming a boxcar light curve, yielding an SNR of the form $(\delta /\sigma _0) \sqrt{D}$. In this work, a general framework is outlined for calculating the SNR of any analytic function and it is applied to the specific case of a trapezoidal transit as a demonstration. By refining the approximation from boxcar to trapezoid, an improved SNR equation is obtained that takes the form $(\delta /\sigma _0) \sqrt{(T_{14}+2T_{23})/3}$. A solution is also derived for the case of a trapezoid convolved with a top-hat, corresponding to observations with finite integration time, where it is proved that SNR is a monotonically decreasing function of integration time. As a rule of thumb, integration times exceeding T14/3 lead to a 10 per cent loss in SNR. This work establishes that the boxcar transit is approximate and it is argued that efforts to calculate accurate completeness maps or occurrence rate statistics should either use the refined expression, or even better numerically solve for the SNR of a more physically complete transit model.