Representations of Precipitation Diurnal Cycle in the Amazon as Simulated by Observationally Constrained Cloud‐System Resolving and Global Climate Models

The ability of an observationally‐constrained cloud‐system resolving model (Weather Research and Forecasting; WRF, 4‐km grid spacing) and a global climate model (Energy Exascale Earth System Model; E3SM, 1‐degree grid spacing) to represent the precipitation diurnal cycle over the Amazon basin during the 2014 wet season is assessed. The WRF model coupled with a 3‐D variational data assimilation scheme reproduces the spatial variability of the precipitation diurnal cycle over the basin and the lifecycle of westward propagating MCSs initiated by the coastal sea‐breeze front. In contrast, a single morning peak in rainfall is produced by E3SM for simulations despite the nudging of large‐scale winds toward global reanalysis, indicating precipitation in E3SM is largely controlled by local convection associated with diurnal heating. The role of propagating MCS on the environment are discussed by using a multivariate perturbation analysis. We also find that the advection of moisture perturbations from ocean to inland regions have a higher correlation with the occurrence of MCSs in the Amazon than the intensity of colder air intrusion associated with sea breezes along the coast. Moreover, the presence of large cold pools over the central Amazon basin are responsible for the maintenance of propagating deep convection. Plain Language Summary The Amazon basin in South America is one of the regions over land that has the highest occurrence of large‐size and deep cloud systems (also called “Mescoscale Convective System” [MCS]). Since they have a wide coverage and produce much heavier rainfall than the other types of cloud, the regional climate and even the earth system are tied closely with their behaviors. However, current global atmospheric models are unable to reproduce realistic diurnal variation of precipitation in the Amazon and the poor representation of those MCSs is responsible for the deficiency. We use various observations as the reference to understand how accurate the physical processes related to MCS are represented by both the cloud‐system resolving (higher‐resolution) and global climate (lower‐resolution) models. The results show the diurnal variation of local precipitation in the basin is mostly reproduced by cloud‐system resolving model but not the global climate model, because the propagating MCSs and many related processes can only be simulated by using higher‐resolution model. We also found the advection of moisture perturbation from ocean to inland