A change in measured transit depth does not necessarily mean an exoplanet changed size. Depth is the blocked light divided by a measured baseline. Both quantities depend on the star, neighboring light, wavelength and analysis. The first task is to establish whether two observations measure the same thing.
Compare like with like
Check passband, aperture, contamination correction, exposure time and the definition of depth. The deepest data point, a fitted radius ratio squared and a limb-darkened model’s mid-transit flux loss are not always identical quantities.
Observations at different wavelengths can legitimately yield different effective radii because an atmosphere absorbs and scatters differently. That is the basis of transmission spectroscopy. It is not automatically evidence for time-variable atmospheric conditions.
Dilution can make an unchanged planet look smaller
If a neighboring star contributes extra flux, the observed depth is δobs = δhost Fhost/(Fhost + Fneighbor), assuming the neighbor is not involved in the transit. For an illustrative true depth of 1% and neighbor flux equal to 20% of the host’s flux, the measured depth is 0.833%.
Changing the extraction aperture or instrument can change that contamination. The planet did not shrink. Compare high-resolution imaging, aperture tests and the instrument’s contamination treatment before interpreting depth differences physically.
The baseline star can change
Uncrossed dark spots can deepen a normalized transit; crossed spots can add bumps and bias a fit. Bright regions can act differently. The multiwavelength stellar-activity study demonstrates why the passband matters. Read the dedicated starspot explanation for the underlying flux calculation.
Rotational monitoring helps, but a low-amplitude rotational signal does not establish a uniform stellar surface. Surface patterns may be nearly symmetric or remain outside the transit chord.
Geometry and fitting can trade off
A change in impact parameter can move the transit across a different stellar brightness profile. Near-grazing events are particularly sensitive to geometry. Limb darkening, radius ratio, duration and baseline trends can be correlated in the fit.
Incomplete ingress or egress and different treatments of correlated noise can consequently produce apparently significant depth changes. Test whether the effect remains with a shared physical model, realistic exposure integration and alternative reasonable detrending choices.
When a changing planet becomes plausible
Variable clouds, escaping dusty material or other structures may be relevant for particular systems. Require repeatability, wavelength information and a physical model that explains the event shape as well as its depth. Ordinary broadband depth variation does not identify a molecule or prove weather.
The study of transit radii from ingress and egress durations explores separating geometrical radius information from brightness contamination. It also illustrates why contact geometry and depth should be considered together.
Next compare transit asymmetry and duration changes. A proposed explanation should account for all measured changes, including those that remain absent.
Revised October 26, 2026. The dilution calculation is illustrative.