PNNL

Abstract

The severe impacts of rain-on-snow (ROS) extreme events have been widely recognized and studied. However, unlike hydrological processes, flood inundation dynamics during ROS events remain under-investigated, despite their importance for flood risk assessment and management. We diagnosed and documented heterogeneous, sub-kilometer spatio-temporal dynamics of flood inundation during the 2017 California ROS events, simulated by a 2-dimensional hydrodynamic model, River Dynamic Core (RDycore). RDycore shows good performance in capturing fine-scale flood inundation dynamics against gauge measurements (with a median correlation coefficient of 0.81) and satellite observations. A Random Forest-based analysis shows that topography plays a critical role in shaping maximum water height, accounting for 34% of its spatial variability, while runoff volume explains 22%. Even with small runoff, regions with flat slopes or elevation below 1500 m can experience high flooded area ratios exceeding 10%. Our findings offer an explicit and accurate picture of when and where ROS flooding could occur, valuable for risk assessment and infrastructure planning.