Cumulative disturbances to assess forest degradation using spectral unmixing in the northeastern Amazon

Abstract Question Tropical forests are subject to disturbances by logging, gathering of fuelwood, and fires. Can degradation trajectories (i.e. cumulative disturbances events over a period of timer) be identified using remote‐sensing Landsat time series? Location Paragominas (Pará, Brazil), a municipality covering 19,395 km² in the north‐eastern Amazon. Methods We used Landsat annual imagery from 2000 to 2015 and spectral mixture analysis to derive time series of the fraction of soil (S), active photosynthetic vegetation ( PV ), and non‐photosynthetic vegetation (senescent) ( NPV ) indicators. Results The NPV values over a 16‐year period revealed five different degradation trajectories (i.e., cumulative disturbances in space and over time): undisturbed forest, selectively logged forest (with a management plan), overlogged forest (no management plan), overlogged forest (charcoal production) and burned forest. The variance of NPV calculated per pixel over the same period is useful to map forest degradation over Paragominas municipality, highlighting the role of disturbance factors (logging, fuelwood gathering and fire). Conclusions The fractional cover of NPV obtained from spectral mixture analysis can be used to differentiate degradation trajectories and to map forest degradation.