Ground-Based Follow-up Analysis of TOI-3669.01 from the GMU Observatory (NASA Data Science Analysis 2025 Program)

This paper presents a follow-up evaluation of TESS exoplanet candidate TOI-3669.01 from a single night of ground-based observation at George Mason University Observatory (GMUO) on 2023-09-06 through the R filter (Ricker et al. 2015). When it came to the analysis of TOI-3669’s ephemeris, AstroImageJ (AIJ) served multiple important purposes; these include data reduction, plate-solving, and differential aperture photometry (Collins et al. 2017). During the detrending of our target’s flux data, the parameters used include AIRMASS, FWHM, sky level per pixel, and centroid shifts. At an effective cadence of ∼ 147 s (exposure + overhead), the sequence comprises 156 frames over ∼ 0.265 d (≈ 382 min). A phase-consistent transit-like depression with AIJ best-fit depth ∼ 19.6 ppt (≈ 1.96%) and total duration t14 ≈ 0.103 d is present within the predicted event window (Mandel & Agol 2002). To assess contamination from other stars, we used AIJ’s NEB diagnostics and an RMS–Δmag diagram, and inspected centroids within 10’ of the target; given the noise levels and measured depths, none of the neighboring sources showed coincident eclipses that could be the cause for the signal at the target’s position (Collins et al. 2017). This suggests that the target is a source of the dimming that was observed. To better contextualize the quality and achievable sensitivity of a single-night dataset, an injection–recovery analysis was run on the reduced light curve (e.g., Christiansen et al. 2015; Christiansen et al. 2017; Petigura et al. 2013), implemented with our Python workflow TransitBench (Tripathi 2025). A permissive (“sensitive”) search profile was used, with the out-of-transit-replacement method recovering most injected transits over depths ∼ 0.092–0.12 at a global TPR of ∼ 0.89. The other method of injection-recovery analysis, a prewhitening search, didn’t perform well with a TPR of 0, leading to its exclusion from consideration in this paper and study. Its underperformance is likely due to the abundance of red noise in this dataset. When looking at the generated robustness curves, a trend is visible: as adversarial perturbations are added, the rate of recovery for the synthetic injections decreases, which is typical for short, red-noise–dominated datasets (Pont, Zucker & Queloz 2006); the curves were produced with TransitBench (Tripathi 2025). This study does not confirm that TOI-3669.01 is an exoplanet, but it remains a viable candidate. In future studies, multi-night and multi-filter follow-up analysis is suggested to verify TOI-3669.01’s status as an exoplanet.