Consideration of carbon credits to developing countries for averted deforestation is high on the policy agenda. The underpinnings for such policies rest on technical capabilities to estimate and monitor carbon emissions. Changing dynamics of deforestation in some parts of the tropics towards mechanized, large-scale production challenges current methods to estimate emissions. The presentation will discuss scientific uncertainties in estimating deforestation emissions and new approaches to modeling carbon fluxes using a variety of remotely-sensed data inputs. The presentation will focus on two spatial scales, a state-level effort to improve modeling capabilities to estimate deforestation fluxes and a pan-tropical effort to identify hotspots contributing disproportionately high emissions.