Integrating the Fokker-Planck Equation (FPE) subgrid snow model into the Rapid Refresh (RAP) weather forecast system for considering snow heterogeneity in surface-atmosphere interactions

Subgrid variability of snow is still a challenge in weather prediction models while its impact is pretty significant. The Fokker-Planck Equation (FPE) subgrid snow model is a physically-based stochastic model that was developed for simulating the subgrid variability of snow. Offline tests of the FPE subgrid snow model coupled to Rapid Update Cycle (RUC) Land-Surface Model (LSM), which has named RUC-FPE, have provided very promising results. In this presentation, the FPE snow model has been tested in the fully-coupled mode with the NOAA Rapid Refresh (RAP) weather forecast system. The RUC-FPE snow model mainly changes the algorithm for calculating snow cover fraction in the RAP model. Specifically, snow cover fraction in RAP is calculated through a linear depletion curve method, while the FPE snow model calculates snow cover fraction by considering snow redistribution and snowmelt. Offline tests have shown that FPE snow model can capture gradual decrease of snow cover fraction much better than the depletion curve method. Integrating the FPE snow model into the surface component of the RAP model is evaluated and analyzed in this study. The simulated horizontal fields (i.e. snow distribution, 2-m temperature, 10-m wind speed, 2-m relative humidity, and 2-m dewpoint) and vertical profiles (i.e. temperature and relative humidity) from the original RAP and RAP with RUC-FPE enhancement are compared. Overall, the results show that performance of RAP is improved by integrating the FPE snow model, and the results also demonstrate the importance of considering snow heterogeneity in surface-atmosphere interactions.