In this paper, we assess the potential of spatial, temporal and spatio-temporal deep learning methods for large-scale classification of land-use following tropical deforestation using dense satellite time series on the pan-tropical scale. Continental models performed better than the pan-tropical model, while overall spatio-temporal models performed better than spatial or temporal ones.
Our interdisciplinary approach revealed the complexities of local direct and indirect DD drivers, and the complementarity of remotely sensed, spatially modelled and locally reported methods for driver identification. Overall, REDD+ interventions were found to be aligned with deforestation drivers.
The need for data on drivers and activities causing forest carbon change have been highlighted as central components in REDD+ readiness efforts. Assessment of direct and indirect drivers on the national level is often lacking or incomplete. This thesis explores the role of remote sensing for monitoring tropical forests for REDD+ in general, and for assessing land use and related carbon emissions linked to drivers of tropical deforestation in particular.
Countries are encouraged to identify drivers of deforestation and forest degradation (DD) in the development of national strategies and action plans for REDD+. In this letter we provide an assessment of proximate drivers of DD by synthesizing empirical data reported by countries as part of their REDD+ readiness activities and scientific literature.