We introduce a novel method tailored for unconstrained multi-view optical satellite photogrammetry in time-varying illumination and reflection conditions. Our approach employs continuous radiance fields to represent surface radiance and albedo based on radiometry principles, integrating both static and transient components for satellite photogrammetry. Additionally, an innovative self-supervised mechanism is introduced to optimize the learning process which leverages dark regions accentuation, transient and static composition, as well as shadow regularization. Evaluations on multi-date WorldView-3 images affirm that our model consistently surpasses the state-of-the-art techniques.