Remotely sensed far-red solar-induced fluorescence (SIF) is increasingly used as a proxy for modeling vegetation photosynthetic activity and gross primary production. However, the apparent top-of-canopy (TOC) SIF signal is strongly affected by non-physiological factors, e.g., canopy structure, leaf and ground or atmospheric optical properties. This work demonstrates different approaches removing these confounding factors and extracting from remotely-sensed TOC SIF observations only the physiologically relevant part in form of normalized fluorescence efficiencies.