Changes in the surface and atmospheric water budget due to projected Amazon deforestation: Lessons from a fully coupled model simulation

International audience ; The Amazon forest has a complex interaction with climate at different spatial and temporal scales. This means that alterations in land use could modify the regional water cycle, including the surface and atmospheric water budget. However, little is known about how these changes occur seasonally and in a spatially distributed manner in the most vulnerable regions, such as the southern Amazon. In this study, the local to regional effects of future Amazon deforestation on the surface and atmospheric water budget components are investigated by twin numerical experiments using the Regional Earth System Model of the 'Institute Pierre Simone Laplace' (RegIPSL) for 19 yr (2001-2019). The results show that significant changes in precipitation and actual evapotranspiration in the southern Amazon (south of 5°S) are associated with surrounding areas with a deforested ratio higher than 40%. During the onset of the wet season (September-November) the largest changes in convective processes are manifested by opposite atmospheric dynamic in adjacent regions (dipole), associated with. This dynamic is associated with wind orientation and the different sizes of the straight corridors of continuous deforestation (pathways). The dipole manifests itself as a suppression of convection in the upwind sector, while convection increases in the downwind sector of the deforestation pathway. For medium-sized deforestation pathways (∼350 km) convection changes are related to dynamic processes (decrease in surface roughness). In large-sized pathways (∼500 km) the mechanisms causing convective changes are combined, dynamic and thermal (increase in surface temperature). In deforested regions there is an average increase of terrestrial water storage dynamics and runoff ∼10 times higher than in non-deforested regions. Furthermore, the atmosphere becomes ∼8 times drier in deforested regions than in non-deforested regions. Our findings indicate a new perspective regarding a comprehensive modeling approach to understand ...