Civil Engineering Departmental Seminar - How well do current reanalysis products reproduce rainforest-initiated wet season onset over the southern Amazon?

How well do current reanalysis products reproduce rainforest-initiated wet season onset over the southern Amazon?

We have recently demonstrated that transpiration is an essential driver of the seasonal cycle of rainfall over the southern Amazon, where the potential for rainforest dieback has been flagged as a potential ’tipping point’ in global climate change. Transpiration by the subtropical rainforest in this region enables an increase of shallow convection that moistens and destabilizes the atmosphere during the initial stages of the dry-to-wet season transition. This shallow convection moisture pump preconditions the atmosphere at the regional scale for a rapid increase in rain-bearing deep convection, which in turn drives moisture convergence and wet season onset several months before the arrival of the Intertropical Convergence Zone. However, both shallow convection and land–atmosphere interactions are difficult to represent in atmospheric models, including the reanalysis products that underpin much of our mechanistic understanding of the atmospheric water cycle. Here, we use a refined budget decomposition to evaluate the dry-to-wet season transition in the southern Amazon as represented by three recent reanalyses: ERA5, MERRA-2, and JRA-3Q. The budgets separately diagnose the roles of resolved dynamical processes, parameterized physical processes, and data assimilation in the evolution of moisture and thermal energy as represented by these three reanalyses. We show that reduced turbulent and advective moistening in MERRA-2 are compensated by large positive assimilation increments in lower tropospheric water vapor, implying systematic errors in physics, dynamics, or both. Although both ERA5 and JRA-3Q have their own idiosyncrasies, these systems plainly produce an earlier, smoother transition to deep convective heating with a vertical structure that better promotes regional-scale convergence, and a reduced reliance on data assimilation. The results establish a valuable process-oriented framework for assessing model representations of rainfall seasonality over the subtropical Amazon, where pressures exerted by climate change and deforestation have greatly eroded drought resilience.