Three noise models retain a positive depth and post-transit flux relation in 108 K2-22 Campaign 1 orbits with linear baselines

We test whether time-correlated flux errors remove the reported relation between core transit depth and flux after transit in K2-22. We use 108 matched Campaign 1 orbits from K2SFF and EVEREST. Noise fits use only samples outside transit. White, white plus exponential, and signed AR(1) models all give positive joint five-record block intervals with linear baselines. The exponential slopes are 0.0766 for K2SFF (interval 0.0306 to 0.1695) and 0.0380 for EVEREST (interval 0.0153 to 0.0731). Resampling refits the noise laws. Later-data baseline scores give no clear improvement over white noise. Quadratic baselines with white noise give intervals that include zero in both products. Thus the result remains conditional on baseline and noise assumptions. It is consistent with dust extinction and scattering but does not confirm that mechanism. The reductions share exposures. No independent instrument confirms the relation. Code, exact source hashes, aggregate grids, figures and separate-scratch reproduction commands are included. No external observations or record-level extracts are bundled.

astronomy · K2-22 · dust transit · time-correlated noise · K2SFF · EVEREST · covariance · baseline sensitivity