Abstract Climatological analyses of moisture sources of precipitation have traditionally relied on reanalyses or models that parameterize convection. Convection‐permitting models (CPMs) are increasingly used in climate studies, as they better represent many precipitation processes than non‐CPMs. We found significant differences in precipitation moisture sources over the Amazon Basin using 1‐year CPM and non‐CPM WRF simulations with moisture tracers. Notably, the CPM estimates that about half of precipitation in the central Andes comes from the Amazon basin; a 20%–30% higher estimate than the non‐CPM. This suggests long‐term CPMs with tracers could improve climatological estimates. However, their high computational cost is prohibitive. To overcome this, we developed a revised 2L‐DRM model that replicates CPM‐with‐tracers estimates at a fraction of the cost, using only standard outputs. We applied this model to South America, analyzing precipitation moisture sources across 15 regions. 2L‐DRM can be used for other regions as continental‐scale CPM climatological simulations become available.
Changhai Liu (Sun,) studied this question.