Impact of Cloud Parameterization on Rainfall Simulations over the South Amazonia in Community Atmospheric Model Version 5 (CAM5)

In this study, we conduct four numerical experiments to investigate the differences between rainfall over the Southern Amazonia (SA) simulated by the Community Atmosphere Model version 5 (CAM5), superparameterized (SP) CAM5 (SPCAM, a multiscale modeling framework model), and two models called CAM5_CLUBB and SPCAM_CLUBB, which are implemented with a higher-order turbulence closure scheme, called Cloud Layers Unified By Binomials (CLUBB). Results show that both SP and CLUBB effectively reduce the low biases of rainfall, mainly during wet season. The CLUBB further reduces low biases of humidity in the low atmosphere, but increases the low biases of shallow clouds. The latter enables more solar heat flux to the surface, leads to stronger convection with more rainfall. SP makes the atmosphere more unstable with more moisture and higher atmospheric temperatures, further generating more deep convection and allowing the growth of more extreme convection. 

SP significantly increases deep convection and precipitation in the afternoon, but it does not reduce the early bias of the diurnal peak of the rainfall; CLUBB, on the other hand, improves the diurnal cycle of rainfall by delaying the afternoon peak time and producing precipitation in the early morning, due to more realistic gradual transition between shallow and deep convection. SP can increase or even overestimate precipitation in the afternoon, but does not significantly delay the timing of precipitation peak.

The relationship between the ambient conditions and development of convection also vary among these models. In reality, increased relative humidity is observed prior to deep convection events. SP appears to enable the SPCAM and SPCAM_CLUBB to capture this phenomenon. In contrast, the CAM5 and CAM5_CLUBB tend to have more precipitation with more solar heating in the low atmosphere.