Channeled spectropolarimeters are optical instruments that measure the spectral dependence of the polarization of light without any mechanically moving parts. An important factor in achieving stable and accurate measurements is the calibration process, especially in dynamic environments where temperature fluctuations or other factors affect the retardance of the components in the polarimeter. In previous research, a self-calibration algorithm that accounts for these variations was developed, without any additional reference measurements. In this paper, we identify an ambiguity in the self-calibration phase, which limits the allowed temperature changes to surprisingly small ranges. We show how to adaptively estimate and correct for the phase ambiguity using a polynomial curve-fitting algorithm, extending the temperature range to virtually all practical scenarios. Lastly, we demonstrate the ability of the modified self-calibration algorithm to provide stable reconstruction of the Stokes vector for a temperature range >40∘C, using an experimental channeled spectropolarimeter.