A Promising Downscaling Strategy for Topographic Heavy Rainfalls over the Asian-Australian Monsoon Region by Leveraging Multi-Scale Moisture Dynamical Control

Conventional downscaling methods are severely challenged in providing future projections for local precipitation extremes over tropical areas with complex terrain. Here, we investigate the multi-scale moisture dynamical control on the topographic heavy rainfalls over Taiwan to construct an alternative downscaling pathway based on crucial physical processes. We identify large-scale regimes of vertically-integrated vapor transport (IVT) with objective machine learning classification and then quantify embedded upstream IVT conditions that drive distinct local rainfall responses over Taiwan, which are verified by semi-realistic large-eddy simulations with realistic topography. The identified environmental control and the resolved convective processes form the physical basis for providing interpretable and high-resolution rainfall projections in a changing climate. In addition to Taiwan, the applicability of this downscaling strategy to four areas in the Asian-Australian monsoon region exhibiting high sensitivity of heavy topographic rainfall to the upstream moisture dynamics is also discussed.