Amazon deforestation fires increase plant productivity through changes in diffuse radiation

Over the past few decades a large increase in carbon storage has been observed in undisturbed forests across Amazonia.[br/] The reason for such a sink is unclear, although many possible mechanisms have been suggested, including changes in temperature, carbon dioxide, precipitation, clouds, and solar radiation. In this work we focus on one such mechanism, namely the increase in plant photosynthesis due to changes in diffuse radiation caused by atmospheric aerosols from large-scale deforestation fires that now occur throughout the Amazon region. We estimate that this mechanism has increased dry season (August-September) net primary productivity (NPP) by up to 30% across wide regions of the Amazon.[br/] We conclude that aerosol from deforestation fires may be responsible for a substantial fraction of the Amazon carbon sink that has been observed. Our approach is based on the combined use of three models: (i) the Global Model of Aerosol Processes (GLOMAP), (ii) the Edwards-Slingo radiation model, and (iii) the UK Met Office JULES land-surface scheme, constrained against in-situ aerosol and radiation observation datasets from several Amazonian sites. A 10 year (1999-2008) GLOMAP simulation using GFED3 biomass burning emissions is first evaluated against aerosol observations, indicating that the model is able to capture the Amazon aerosol seasonality, with enhanced concentrations during the dry season driven by biomass burning. The radiation scheme is then shown to be in good agreement with total and diffuse radiation in-situ observations, the model being able to capture the high total and low diffuse radiation flux in the dry season, as well as the low total and high diffuse radiation flux in the wet season. We then use our modelling framework to quantify the contribution of deforestation fires to diffuse/direct radiation fraction and forest productivity.[br/] We calculate that deforestation fires increase dry season diffuse radiation by up to 60% or 30 Wm-2. Finally, we use the JULES model to show that this increase in diffuse radiation is responsible for a substantial growth in gross primary productivity (GPP), enhancing Amazon-wide dry-season GPP by 5% with local increases of up to 15%. Most of this GPP response results in an increase in NPP, estimated in the dry season at 10% across the Amazon with local increases as large as 30%. This substantial NPP enhancement spatially matches observed increases in forest biomass storage across the Amazon.[br/] We thus suggest that deforestation fires have an important impact on the Amazon carbon budget and attempt to estimate the fraction of the observed forest carbon sink that can be attributed to this mechanism. Change [%] in diffuse radiation due to deforestatio