Numerical representation of turbulent fluxes during the March 2019 Nebraska rain-on-snow event

Abstract Rain-on-snow (ROS) events—during which liquid precipitation falls on an existing surface snowpack—are highly impactful to society, with severe flooding being the primary hazard. ROS events remain a highly challenging hydrometeorological problem in several aspects of land surface model development, pushing the limits of land-atmosphere, snowpack, and runoff modeling. In particular, the representation of turbulent fluxes during these events is critical as energy into the snowpack controls the rate of melt and may impact the magnitude of resulting flooding. In this study, we investigate the representation of these turbulent fluxes in the Weather Research & Forecasting (WRF) model coupled to the Noah-MP land model during a ROS event. For this case study, we use an extreme ROS event which occurred on 12–13 March 2019 across Nebraska, Iowa, and Missouri, resulting in historic flooding and damages. Our WRF simulation of this event was compared with observations from AmeriFlux, snow products, and ERA5 reanalysis fields. While the simulation was able to produce the synoptic dynamics leading up to and during the event, there were notable discrepancies between the observed and modeled turbulent fluxes, suggesting that during ROS events, latent heat flux into the snowpack is underrepresented. Furthermore, analysis of the kilometer-scale WRF simulation run across the CONtiguous United States for 40 years at 4 km resolution (CONUS404) reveals the same underrepresented latent heat fluxes. Simple snowmelt and runoff models, forced with the observed fluxes as well as an experiment with reduced latent heat fluxes, shows that including the latent heat flux melts the snowpack quicker than without it, which has implications for the modeling of flooding in the region.