The Role of Divergent Flow in the Parameterized and Resolved Convective Precipitation over the Amazon Forest

The Amazon tropical forest is a key region for the global-scale biogeochemical cycles and associated climate feedback owing to its abundant rainfall. However, realistically simulating the rainfall over the region remains a challenging task for global models. Our previous study focusing on the Community Atmosphere Model version 5 (CAM5) during the wet season suggested inaccurate divergent circulations over the central-western Amazon leads to biased moisture convergence, which then affect the coupling between the tropospheric moisture and parameterized deep convection. The biased mass divergence is likely a dynamic response of the northeasterly flow to the Guiana Highlands that are marginally resolved by 1° grid. Here we test this scenario by comparing CAM5 simulations with the default finite-volume dynamical core and with the Model for Prediction Across Scales (MPAS), a dynamical core that has different characteristics in mesoscale divergent flow, both run at 1° grids. A convection-permitting (Δx 4km) simulation is also run by using MPAS's regionally refined mesh centered over the Amazon, serving as a reference solution by better resolving the topography and other surface forcings. The simulation period includes GoAmazon's Intensive Operation Period 1 whose observations serve as additional reference. Preliminary results focusing on the divergent circulations, tropospheric moisture, and its influence on the diurnal cycle of deep convection will be presented.