Rectifying Seasonality in the Amazon by Diurnal Cycles

The diurnal and seasonal radiation and water cycles in the Amazon remain improperly modeled in General Circulation Models (GCMs), exhibiting peak evapotranspiration in the wrong season and too early in the day. We here show that those biases are not present in simulations that resolve deep convection but parameterize large-scale dynamics, the opposite of what is done in GCMs. Given the insolation and the domain-averaged temperature profile, our simulations successfully captures both the afternoon precipitation and cloud cover peak and the greater latent heat flux in the dry season than wet. These are both features that GCMs fail to capture, with implications for the correct simulation of evaporation and gross primary production in the Amazon. One of the key findings is that the morning fog layer near the surface in the wet season is crucial in determining the surface energy budget and precipitation. Another finding is that the free-tropospheric temperature profile, warmer in the wet than the dry season (though the opposite is true near the surface), mediates the seasonal transition, allowing precipitation to increase by shallowing the large-scale ascent profile. These simulations pave the way to better understanding the water, energy and carbon cycles over the Amazon forest, as well as to better predicting the impact of deforestation and global warming in the basin.